diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,15613 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "global_step": 2598,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9230769230769234e-07,
+      "loss": 0.748,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.846153846153847e-07,
+      "loss": 0.8339,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.76923076923077e-07,
+      "loss": 0.8677,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.692307692307694e-07,
+      "loss": 0.7595,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.615384615384617e-07,
+      "loss": 0.8411,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.153846153846154e-06,
+      "loss": 0.889,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3461538461538462e-06,
+      "loss": 0.8449,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 0.7679,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7307692307692308e-06,
+      "loss": 0.7627,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9230769230769234e-06,
+      "loss": 0.6547,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1153846153846155e-06,
+      "loss": 0.8574,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 0.7796,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5e-06,
+      "loss": 0.6886,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6923076923076923e-06,
+      "loss": 0.7592,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8846153846153845e-06,
+      "loss": 0.8796,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 0.6729,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2692307692307696e-06,
+      "loss": 0.8391,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4615384615384617e-06,
+      "loss": 0.7579,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.653846153846154e-06,
+      "loss": 0.7533,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 0.763,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0384615384615385e-06,
+      "loss": 0.7617,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.230769230769231e-06,
+      "loss": 0.6282,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.423076923076924e-06,
+      "loss": 0.6849,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 0.622,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.807692307692308e-06,
+      "loss": 0.5601,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5e-06,
+      "loss": 0.6913,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.192307692307693e-06,
+      "loss": 0.6937,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.384615384615385e-06,
+      "loss": 0.6379,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.576923076923077e-06,
+      "loss": 0.7104,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.769230769230769e-06,
+      "loss": 0.6435,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.961538461538462e-06,
+      "loss": 0.6564,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 0.8163,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.3461538461538466e-06,
+      "loss": 0.5748,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.538461538461539e-06,
+      "loss": 0.6662,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.730769230769232e-06,
+      "loss": 0.6704,
+      "step": 35
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 0.7141,
+      "step": 36
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.115384615384616e-06,
+      "loss": 0.6568,
+      "step": 37
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.307692307692308e-06,
+      "loss": 0.6951,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.692,
+      "step": 39
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 0.638,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.884615384615384e-06,
+      "loss": 0.5329,
+      "step": 41
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.076923076923077e-06,
+      "loss": 0.6027,
+      "step": 42
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.26923076923077e-06,
+      "loss": 0.5413,
+      "step": 43
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.461538461538462e-06,
+      "loss": 0.5174,
+      "step": 44
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.653846153846155e-06,
+      "loss": 0.6428,
+      "step": 45
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.846153846153847e-06,
+      "loss": 0.533,
+      "step": 46
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.03846153846154e-06,
+      "loss": 0.6002,
+      "step": 47
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 0.6358,
+      "step": 48
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.423076923076923e-06,
+      "loss": 0.578,
+      "step": 49
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.615384615384616e-06,
+      "loss": 0.6813,
+      "step": 50
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.807692307692308e-06,
+      "loss": 0.7042,
+      "step": 51
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1e-05,
+      "loss": 0.5337,
+      "step": 52
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0192307692307692e-05,
+      "loss": 0.6092,
+      "step": 53
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0384615384615386e-05,
+      "loss": 0.5098,
+      "step": 54
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0576923076923078e-05,
+      "loss": 0.5679,
+      "step": 55
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 0.623,
+      "step": 56
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.0961538461538464e-05,
+      "loss": 0.7148,
+      "step": 57
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1153846153846154e-05,
+      "loss": 0.6725,
+      "step": 58
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1346153846153847e-05,
+      "loss": 0.6691,
+      "step": 59
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1538461538461538e-05,
+      "loss": 0.6184,
+      "step": 60
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1730769230769232e-05,
+      "loss": 0.5759,
+      "step": 61
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1923076923076925e-05,
+      "loss": 0.557,
+      "step": 62
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2115384615384615e-05,
+      "loss": 0.5218,
+      "step": 63
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 0.6424,
+      "step": 64
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.25e-05,
+      "loss": 0.7489,
+      "step": 65
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.2692307692307693e-05,
+      "loss": 0.6603,
+      "step": 66
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.2884615384615386e-05,
+      "loss": 0.6116,
+      "step": 67
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3076923076923078e-05,
+      "loss": 0.5895,
+      "step": 68
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3269230769230769e-05,
+      "loss": 0.6514,
+      "step": 69
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3461538461538463e-05,
+      "loss": 0.4221,
+      "step": 70
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3653846153846156e-05,
+      "loss": 0.4674,
+      "step": 71
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 0.6221,
+      "step": 72
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.403846153846154e-05,
+      "loss": 0.6976,
+      "step": 73
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4230769230769232e-05,
+      "loss": 0.6304,
+      "step": 74
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4423076923076924e-05,
+      "loss": 0.582,
+      "step": 75
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4615384615384615e-05,
+      "loss": 0.5342,
+      "step": 76
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.480769230769231e-05,
+      "loss": 0.5039,
+      "step": 77
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.5408,
+      "step": 78
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5192307692307693e-05,
+      "loss": 0.6092,
+      "step": 79
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 0.6761,
+      "step": 80
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.557692307692308e-05,
+      "loss": 0.6212,
+      "step": 81
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.576923076923077e-05,
+      "loss": 0.5796,
+      "step": 82
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.5961538461538465e-05,
+      "loss": 0.4811,
+      "step": 83
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6153846153846154e-05,
+      "loss": 0.8396,
+      "step": 84
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6346153846153847e-05,
+      "loss": 0.625,
+      "step": 85
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.653846153846154e-05,
+      "loss": 0.7104,
+      "step": 86
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.673076923076923e-05,
+      "loss": 0.603,
+      "step": 87
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 0.505,
+      "step": 88
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7115384615384617e-05,
+      "loss": 0.7497,
+      "step": 89
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.730769230769231e-05,
+      "loss": 0.5523,
+      "step": 90
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 0.6137,
+      "step": 91
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7692307692307694e-05,
+      "loss": 0.5381,
+      "step": 92
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7884615384615387e-05,
+      "loss": 0.4866,
+      "step": 93
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.807692307692308e-05,
+      "loss": 0.5602,
+      "step": 94
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.826923076923077e-05,
+      "loss": 0.4916,
+      "step": 95
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 0.6813,
+      "step": 96
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8653846153846157e-05,
+      "loss": 0.6488,
+      "step": 97
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8846153846153846e-05,
+      "loss": 0.6902,
+      "step": 98
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.903846153846154e-05,
+      "loss": 0.5509,
+      "step": 99
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 0.59,
+      "step": 100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9423076923076924e-05,
+      "loss": 0.7225,
+      "step": 101
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615384615384617e-05,
+      "loss": 0.625,
+      "step": 102
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.980769230769231e-05,
+      "loss": 0.556,
+      "step": 103
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.6211,
+      "step": 104
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999999206628137e-05,
+      "loss": 0.7122,
+      "step": 105
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9999968265138078e-05,
+      "loss": 0.5122,
+      "step": 106
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999992859660788e-05,
+      "loss": 0.5027,
+      "step": 107
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9999873060753727e-05,
+      "loss": 0.6038,
+      "step": 108
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999980165766374e-05,
+      "loss": 0.6637,
+      "step": 109
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999714387451213e-05,
+      "loss": 0.6165,
+      "step": 110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999961125025462e-05,
+      "loss": 0.6255,
+      "step": 111
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999492246237623e-05,
+      "loss": 0.6604,
+      "step": 112
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999357375589045e-05,
+      "loss": 0.6717,
+      "step": 113
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999206638522888e-05,
+      "loss": 0.6371,
+      "step": 114
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999040035278335e-05,
+      "loss": 0.6568,
+      "step": 115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999885756611975e-05,
+      "loss": 0.6704,
+      "step": 116
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9998659231336655e-05,
+      "loss": 0.5779,
+      "step": 117
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999844503124376e-05,
+      "loss": 0.6061,
+      "step": 118
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9998214966180948e-05,
+      "loss": 0.5594,
+      "step": 119
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999796903651327e-05,
+      "loss": 0.6106,
+      "step": 120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999770724263095e-05,
+      "loss": 0.4927,
+      "step": 121
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9997429584949395e-05,
+      "loss": 0.6797,
+      "step": 122
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9997136063909175e-05,
+      "loss": 0.434,
+      "step": 123
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9996826679976033e-05,
+      "loss": 0.6693,
+      "step": 124
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999650143364088e-05,
+      "loss": 0.6177,
+      "step": 125
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9996160325419796e-05,
+      "loss": 0.6113,
+      "step": 126
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999580335585404e-05,
+      "loss": 0.5448,
+      "step": 127
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9995430525510025e-05,
+      "loss": 0.6934,
+      "step": 128
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999504183497934e-05,
+      "loss": 0.6501,
+      "step": 129
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9994637284878733e-05,
+      "loss": 0.5656,
+      "step": 130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9994216875850126e-05,
+      "loss": 0.479,
+      "step": 131
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99937806085606e-05,
+      "loss": 0.7427,
+      "step": 132
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9993328483702393e-05,
+      "loss": 0.5639,
+      "step": 133
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9992860501992924e-05,
+      "loss": 0.5327,
+      "step": 134
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999237666417475e-05,
+      "loss": 0.5399,
+      "step": 135
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9991876971015598e-05,
+      "loss": 0.6644,
+      "step": 136
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9991361423308357e-05,
+      "loss": 0.5826,
+      "step": 137
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999083002187107e-05,
+      "loss": 0.5963,
+      "step": 138
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9990282767546926e-05,
+      "loss": 0.6001,
+      "step": 139
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9989719661204288e-05,
+      "loss": 0.6418,
+      "step": 140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9989140703736657e-05,
+      "loss": 0.4913,
+      "step": 141
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9988545896062687e-05,
+      "loss": 0.5914,
+      "step": 142
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.998793523912619e-05,
+      "loss": 0.5113,
+      "step": 143
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.998730873389612e-05,
+      "loss": 0.5747,
+      "step": 144
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9986666381366583e-05,
+      "loss": 0.6911,
+      "step": 145
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9986008182556827e-05,
+      "loss": 0.6221,
+      "step": 146
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.998533413851124e-05,
+      "loss": 0.5853,
+      "step": 147
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.998464425029936e-05,
+      "loss": 0.584,
+      "step": 148
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9983938519015868e-05,
+      "loss": 0.6063,
+      "step": 149
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.998321694578057e-05,
+      "loss": 0.5902,
+      "step": 150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9982479531738424e-05,
+      "loss": 0.528,
+      "step": 151
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9981726278059514e-05,
+      "loss": 0.6881,
+      "step": 152
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.998095718593906e-05,
+      "loss": 0.5938,
+      "step": 153
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.998017225659742e-05,
+      "loss": 0.4449,
+      "step": 154
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.997937149128007e-05,
+      "loss": 0.7116,
+      "step": 155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9978554891257618e-05,
+      "loss": 0.5021,
+      "step": 156
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99777224578258e-05,
+      "loss": 0.6009,
+      "step": 157
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9976874192305482e-05,
+      "loss": 0.644,
+      "step": 158
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9976010096042634e-05,
+      "loss": 0.6135,
+      "step": 159
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9975130170408362e-05,
+      "loss": 0.6854,
+      "step": 160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9974234416798878e-05,
+      "loss": 0.6012,
+      "step": 161
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9973322836635517e-05,
+      "loss": 0.5954,
+      "step": 162
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9972395431364724e-05,
+      "loss": 0.6506,
+      "step": 163
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9971452202458048e-05,
+      "loss": 0.5623,
+      "step": 164
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9970493151412153e-05,
+      "loss": 0.6396,
+      "step": 165
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.996951827974881e-05,
+      "loss": 0.7226,
+      "step": 166
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.996852758901489e-05,
+      "loss": 0.6148,
+      "step": 167
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9967521080782365e-05,
+      "loss": 0.5817,
+      "step": 168
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9966498756648305e-05,
+      "loss": 0.5693,
+      "step": 169
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9965460618234875e-05,
+      "loss": 0.4865,
+      "step": 170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9964406667189338e-05,
+      "loss": 0.5922,
+      "step": 171
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9963336905184044e-05,
+      "loss": 0.5842,
+      "step": 172
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9962251333916423e-05,
+      "loss": 0.4594,
+      "step": 173
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.996114995510901e-05,
+      "loss": 0.6773,
+      "step": 174
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9960032770509405e-05,
+      "loss": 0.5682,
+      "step": 175
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9958899781890296e-05,
+      "loss": 0.6948,
+      "step": 176
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9957750991049443e-05,
+      "loss": 0.5502,
+      "step": 177
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9956586399809685e-05,
+      "loss": 0.6749,
+      "step": 178
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9955406010018928e-05,
+      "loss": 0.4476,
+      "step": 179
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.995420982355015e-05,
+      "loss": 0.4718,
+      "step": 180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9952997842301386e-05,
+      "loss": 0.5568,
+      "step": 181
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.995177006819575e-05,
+      "loss": 0.561,
+      "step": 182
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99505265031814e-05,
+      "loss": 0.4466,
+      "step": 183
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.994926714923155e-05,
+      "loss": 0.6202,
+      "step": 184
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9947992008344478e-05,
+      "loss": 0.6015,
+      "step": 185
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9946701082543506e-05,
+      "loss": 0.5904,
+      "step": 186
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9945394373877e-05,
+      "loss": 0.5681,
+      "step": 187
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.994407188441837e-05,
+      "loss": 0.5653,
+      "step": 188
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9942733616266076e-05,
+      "loss": 0.5525,
+      "step": 189
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9941379571543597e-05,
+      "loss": 0.6233,
+      "step": 190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9940009752399462e-05,
+      "loss": 0.5963,
+      "step": 191
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9938624161007216e-05,
+      "loss": 0.587,
+      "step": 192
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9937222799565444e-05,
+      "loss": 0.5863,
+      "step": 193
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9935805670297744e-05,
+      "loss": 0.8203,
+      "step": 194
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9934372775452738e-05,
+      "loss": 0.6584,
+      "step": 195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.993292411730406e-05,
+      "loss": 0.5733,
+      "step": 196
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9931459698150368e-05,
+      "loss": 0.7675,
+      "step": 197
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992997952031531e-05,
+      "loss": 0.4997,
+      "step": 198
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9928483586147553e-05,
+      "loss": 0.5405,
+      "step": 199
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992697189802076e-05,
+      "loss": 0.5453,
+      "step": 200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9925444458333598e-05,
+      "loss": 0.5327,
+      "step": 201
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9923901269509712e-05,
+      "loss": 0.5977,
+      "step": 202
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9922342333997758e-05,
+      "loss": 0.5239,
+      "step": 203
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.992076765427136e-05,
+      "loss": 0.6603,
+      "step": 204
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.991917723282913e-05,
+      "loss": 0.5744,
+      "step": 205
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.991757107219467e-05,
+      "loss": 0.5967,
+      "step": 206
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9915949174916533e-05,
+      "loss": 0.582,
+      "step": 207
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.991431154356826e-05,
+      "loss": 0.6459,
+      "step": 208
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.991265818074835e-05,
+      "loss": 0.465,
+      "step": 209
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.991098908908027e-05,
+      "loss": 0.6492,
+      "step": 210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.990930427121244e-05,
+      "loss": 0.6089,
+      "step": 211
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9907603729818228e-05,
+      "loss": 0.5505,
+      "step": 212
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9905887467595963e-05,
+      "loss": 0.6547,
+      "step": 213
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9904155487268912e-05,
+      "loss": 0.5725,
+      "step": 214
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9902407791585286e-05,
+      "loss": 0.6099,
+      "step": 215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.990064438331823e-05,
+      "loss": 0.5776,
+      "step": 216
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9898865265265817e-05,
+      "loss": 0.5793,
+      "step": 217
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.989707044025105e-05,
+      "loss": 0.4843,
+      "step": 218
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9895259911121866e-05,
+      "loss": 0.6006,
+      "step": 219
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9893433680751105e-05,
+      "loss": 0.5583,
+      "step": 220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.989159175203652e-05,
+      "loss": 0.5695,
+      "step": 221
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9889734127900793e-05,
+      "loss": 0.5124,
+      "step": 222
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9887860811291488e-05,
+      "loss": 0.5529,
+      "step": 223
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9885971805181083e-05,
+      "loss": 0.6178,
+      "step": 224
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.988406711256694e-05,
+      "loss": 0.5603,
+      "step": 225
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9882146736471325e-05,
+      "loss": 0.6163,
+      "step": 226
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.988021067994138e-05,
+      "loss": 0.6368,
+      "step": 227
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.987825894604913e-05,
+      "loss": 0.6269,
+      "step": 228
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9876291537891482e-05,
+      "loss": 0.5428,
+      "step": 229
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9874308458590203e-05,
+      "loss": 0.5585,
+      "step": 230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9872309711291927e-05,
+      "loss": 0.7046,
+      "step": 231
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9870295299168162e-05,
+      "loss": 0.4315,
+      "step": 232
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9868265225415263e-05,
+      "loss": 0.5832,
+      "step": 233
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9866219493254433e-05,
+      "loss": 0.521,
+      "step": 234
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9864158105931726e-05,
+      "loss": 0.5408,
+      "step": 235
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.986208106671804e-05,
+      "loss": 0.6272,
+      "step": 236
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9859988378909096e-05,
+      "loss": 0.6248,
+      "step": 237
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.985788004582546e-05,
+      "loss": 0.6097,
+      "step": 238
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9855756070812514e-05,
+      "loss": 0.6635,
+      "step": 239
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.985361645724046e-05,
+      "loss": 0.514,
+      "step": 240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9851461208504317e-05,
+      "loss": 0.6158,
+      "step": 241
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9849290328023913e-05,
+      "loss": 0.5011,
+      "step": 242
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.984710381924388e-05,
+      "loss": 0.6766,
+      "step": 243
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9844901685633648e-05,
+      "loss": 0.5454,
+      "step": 244
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9842683930687435e-05,
+      "loss": 0.5607,
+      "step": 245
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9840450557924255e-05,
+      "loss": 0.4669,
+      "step": 246
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9838201570887893e-05,
+      "loss": 0.5869,
+      "step": 247
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9835936973146916e-05,
+      "loss": 0.5029,
+      "step": 248
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.983365676829466e-05,
+      "loss": 0.5453,
+      "step": 249
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9831360959949234e-05,
+      "loss": 0.5316,
+      "step": 250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9829049551753484e-05,
+      "loss": 0.592,
+      "step": 251
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9826722547375033e-05,
+      "loss": 0.406,
+      "step": 252
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9824379950506233e-05,
+      "loss": 0.4869,
+      "step": 253
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9822021764864194e-05,
+      "loss": 0.5129,
+      "step": 254
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9819647994190744e-05,
+      "loss": 0.5895,
+      "step": 255
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.981725864225245e-05,
+      "loss": 0.4573,
+      "step": 256
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.98148537128406e-05,
+      "loss": 0.5952,
+      "step": 257
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.981243320977121e-05,
+      "loss": 0.5188,
+      "step": 258
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.980999713688499e-05,
+      "loss": 0.6525,
+      "step": 259
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9807545498047363e-05,
+      "loss": 0.5981,
+      "step": 260
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9805078297148453e-05,
+      "loss": 0.7386,
+      "step": 261
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9802595538103076e-05,
+      "loss": 0.5061,
+      "step": 262
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9800097224850736e-05,
+      "loss": 0.6735,
+      "step": 263
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.979758336135561e-05,
+      "loss": 0.5168,
+      "step": 264
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9795053951606565e-05,
+      "loss": 0.5636,
+      "step": 265
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9792508999617116e-05,
+      "loss": 0.5356,
+      "step": 266
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9789948509425455e-05,
+      "loss": 0.7625,
+      "step": 267
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9787372485094424e-05,
+      "loss": 0.6211,
+      "step": 268
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9784780930711514e-05,
+      "loss": 0.4691,
+      "step": 269
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9782173850388852e-05,
+      "loss": 0.544,
+      "step": 270
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.977955124826321e-05,
+      "loss": 0.6195,
+      "step": 271
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9776913128495988e-05,
+      "loss": 0.5157,
+      "step": 272
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.97742594952732e-05,
+      "loss": 0.6447,
+      "step": 273
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.977159035280549e-05,
+      "loss": 0.5863,
+      "step": 274
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9768905705328096e-05,
+      "loss": 0.5714,
+      "step": 275
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.976620555710087e-05,
+      "loss": 0.6109,
+      "step": 276
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.976348991240825e-05,
+      "loss": 0.6436,
+      "step": 277
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9760758775559275e-05,
+      "loss": 0.4886,
+      "step": 278
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.975801215088755e-05,
+      "loss": 0.482,
+      "step": 279
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9755250042751278e-05,
+      "loss": 0.5891,
+      "step": 280
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9752472455533202e-05,
+      "loss": 0.6785,
+      "step": 281
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9749679393640653e-05,
+      "loss": 0.6311,
+      "step": 282
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9746870861505495e-05,
+      "loss": 0.5594,
+      "step": 283
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.974404686358416e-05,
+      "loss": 0.6605,
+      "step": 284
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9741207404357597e-05,
+      "loss": 0.4592,
+      "step": 285
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9738352488331305e-05,
+      "loss": 0.5799,
+      "step": 286
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9735482120035304e-05,
+      "loss": 0.6562,
+      "step": 287
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9732596304024137e-05,
+      "loss": 0.5513,
+      "step": 288
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9729695044876847e-05,
+      "loss": 0.6825,
+      "step": 289
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9726778347196995e-05,
+      "loss": 0.6431,
+      "step": 290
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.972384621561263e-05,
+      "loss": 0.5716,
+      "step": 291
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9720898654776293e-05,
+      "loss": 0.54,
+      "step": 292
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9717935669365007e-05,
+      "loss": 0.5573,
+      "step": 293
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.971495726408027e-05,
+      "loss": 0.5687,
+      "step": 294
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9711963443648057e-05,
+      "loss": 0.4775,
+      "step": 295
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.970895421281878e-05,
+      "loss": 0.5934,
+      "step": 296
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9705929576367326e-05,
+      "loss": 0.6257,
+      "step": 297
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9702889539093015e-05,
+      "loss": 0.5463,
+      "step": 298
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.969983410581961e-05,
+      "loss": 0.5992,
+      "step": 299
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9696763281395295e-05,
+      "loss": 0.6673,
+      "step": 300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9693677070692687e-05,
+      "loss": 0.6378,
+      "step": 301
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9690575478608807e-05,
+      "loss": 0.5191,
+      "step": 302
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9687458510065086e-05,
+      "loss": 0.5353,
+      "step": 303
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9684326170007365e-05,
+      "loss": 0.6113,
+      "step": 304
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.968117846340585e-05,
+      "loss": 0.7206,
+      "step": 305
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9678015395255153e-05,
+      "loss": 0.5749,
+      "step": 306
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9674836970574253e-05,
+      "loss": 0.6778,
+      "step": 307
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9671643194406497e-05,
+      "loss": 0.5129,
+      "step": 308
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.966843407181958e-05,
+      "loss": 0.7014,
+      "step": 309
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9665209607905565e-05,
+      "loss": 0.5035,
+      "step": 310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.966196980778085e-05,
+      "loss": 0.6775,
+      "step": 311
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.965871467658616e-05,
+      "loss": 0.5319,
+      "step": 312
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9655444219486567e-05,
+      "loss": 0.5911,
+      "step": 313
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9652158441671435e-05,
+      "loss": 0.621,
+      "step": 314
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.964885734835446e-05,
+      "loss": 0.4829,
+      "step": 315
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9645540944773627e-05,
+      "loss": 0.5641,
+      "step": 316
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.964220923619122e-05,
+      "loss": 0.6305,
+      "step": 317
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9638862227893807e-05,
+      "loss": 0.6806,
+      "step": 318
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.963549992519223e-05,
+      "loss": 0.5525,
+      "step": 319
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9632122333421606e-05,
+      "loss": 0.4781,
+      "step": 320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9628729457941305e-05,
+      "loss": 0.5798,
+      "step": 321
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.962532130413495e-05,
+      "loss": 0.6997,
+      "step": 322
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.962189787741041e-05,
+      "loss": 0.7998,
+      "step": 323
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9618459183199782e-05,
+      "loss": 0.5827,
+      "step": 324
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.96150052269594e-05,
+      "loss": 0.5808,
+      "step": 325
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.96115360141698e-05,
+      "loss": 0.6081,
+      "step": 326
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9608051550335733e-05,
+      "loss": 0.5369,
+      "step": 327
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9604551840986157e-05,
+      "loss": 0.5491,
+      "step": 328
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.960103689167421e-05,
+      "loss": 0.5336,
+      "step": 329
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.959750670797721e-05,
+      "loss": 0.5603,
+      "step": 330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9593961295496665e-05,
+      "loss": 0.4854,
+      "step": 331
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.959040065985823e-05,
+      "loss": 0.5343,
+      "step": 332
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9586824806711722e-05,
+      "loss": 0.736,
+      "step": 333
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.95832337417311e-05,
+      "loss": 0.615,
+      "step": 334
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9579627470614472e-05,
+      "loss": 0.3951,
+      "step": 335
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.957600599908406e-05,
+      "loss": 0.4331,
+      "step": 336
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9572369332886213e-05,
+      "loss": 0.4932,
+      "step": 337
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9568717477791388e-05,
+      "loss": 0.457,
+      "step": 338
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.956505043959414e-05,
+      "loss": 0.6101,
+      "step": 339
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9561368224113125e-05,
+      "loss": 0.5147,
+      "step": 340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.955767083719107e-05,
+      "loss": 0.6134,
+      "step": 341
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9553958284694788e-05,
+      "loss": 0.4833,
+      "step": 342
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9550230572515136e-05,
+      "loss": 0.4745,
+      "step": 343
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.954648770656705e-05,
+      "loss": 0.6042,
+      "step": 344
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9542729692789492e-05,
+      "loss": 0.5469,
+      "step": 345
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.953895653714547e-05,
+      "loss": 0.7565,
+      "step": 346
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.953516824562201e-05,
+      "loss": 0.3685,
+      "step": 347
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9531364824230165e-05,
+      "loss": 0.575,
+      "step": 348
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.952754627900499e-05,
+      "loss": 0.6169,
+      "step": 349
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9523712616005538e-05,
+      "loss": 0.496,
+      "step": 350
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.951986384131485e-05,
+      "loss": 0.5976,
+      "step": 351
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9515999961039934e-05,
+      "loss": 0.5788,
+      "step": 352
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9512120981311797e-05,
+      "loss": 0.5953,
+      "step": 353
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9508226908285368e-05,
+      "loss": 0.5223,
+      "step": 354
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9504317748139556e-05,
+      "loss": 0.5071,
+      "step": 355
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.950039350707719e-05,
+      "loss": 0.6289,
+      "step": 356
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9496454191325034e-05,
+      "loss": 0.585,
+      "step": 357
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9492499807133773e-05,
+      "loss": 0.7187,
+      "step": 358
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9488530360778007e-05,
+      "loss": 0.5473,
+      "step": 359
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9484545858556223e-05,
+      "loss": 0.5617,
+      "step": 360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9480546306790807e-05,
+      "loss": 0.6398,
+      "step": 361
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9476531711828027e-05,
+      "loss": 0.5573,
+      "step": 362
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.947250208003801e-05,
+      "loss": 0.4263,
+      "step": 363
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9468457417814753e-05,
+      "loss": 0.5759,
+      "step": 364
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9464397731576093e-05,
+      "loss": 0.5269,
+      "step": 365
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.946032302776372e-05,
+      "loss": 0.3482,
+      "step": 366
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9456233312843144e-05,
+      "loss": 0.6623,
+      "step": 367
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9452128593303685e-05,
+      "loss": 0.6339,
+      "step": 368
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.944800887565849e-05,
+      "loss": 0.5919,
+      "step": 369
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.944387416644449e-05,
+      "loss": 0.6489,
+      "step": 370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9439724472222408e-05,
+      "loss": 0.6172,
+      "step": 371
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9435559799576753e-05,
+      "loss": 0.5124,
+      "step": 372
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9431380155115785e-05,
+      "loss": 0.5285,
+      "step": 373
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9427185545471537e-05,
+      "loss": 0.4223,
+      "step": 374
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9422975977299772e-05,
+      "loss": 0.5386,
+      "step": 375
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9418751457279997e-05,
+      "loss": 0.5656,
+      "step": 376
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9414511992115446e-05,
+      "loss": 0.5889,
+      "step": 377
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9410257588533063e-05,
+      "loss": 0.7145,
+      "step": 378
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9405988253283492e-05,
+      "loss": 0.4621,
+      "step": 379
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9401703993141078e-05,
+      "loss": 0.6924,
+      "step": 380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9397404814903845e-05,
+      "loss": 0.5285,
+      "step": 381
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.939309072539348e-05,
+      "loss": 0.5516,
+      "step": 382
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.938876173145535e-05,
+      "loss": 0.5792,
+      "step": 383
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9384417839958443e-05,
+      "loss": 0.4483,
+      "step": 384
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.938005905779542e-05,
+      "loss": 0.5301,
+      "step": 385
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.937568539188253e-05,
+      "loss": 0.6716,
+      "step": 386
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.937129684915968e-05,
+      "loss": 0.4695,
+      "step": 387
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.936689343659035e-05,
+      "loss": 0.5291,
+      "step": 388
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.936247516116163e-05,
+      "loss": 0.7493,
+      "step": 389
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.935804202988419e-05,
+      "loss": 0.5213,
+      "step": 390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.935359404979228e-05,
+      "loss": 0.561,
+      "step": 391
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9349131227943697e-05,
+      "loss": 0.7418,
+      "step": 392
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.93446535714198e-05,
+      "loss": 0.6388,
+      "step": 393
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9340161087325483e-05,
+      "loss": 0.622,
+      "step": 394
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.933565378278916e-05,
+      "loss": 0.544,
+      "step": 395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9331131664962775e-05,
+      "loss": 0.4293,
+      "step": 396
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9326594741021765e-05,
+      "loss": 0.4307,
+      "step": 397
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9322043018165072e-05,
+      "loss": 0.4072,
+      "step": 398
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9317476503615108e-05,
+      "loss": 0.6045,
+      "step": 399
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.931289520461776e-05,
+      "loss": 0.5462,
+      "step": 400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9308299128442385e-05,
+      "loss": 0.5631,
+      "step": 401
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.930368828238177e-05,
+      "loss": 0.5792,
+      "step": 402
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.929906267375215e-05,
+      "loss": 0.4283,
+      "step": 403
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9294422309893177e-05,
+      "loss": 0.6184,
+      "step": 404
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9289767198167918e-05,
+      "loss": 0.5928,
+      "step": 405
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9285097345962843e-05,
+      "loss": 0.5778,
+      "step": 406
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9280412760687817e-05,
+      "loss": 0.575,
+      "step": 407
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.927571344977607e-05,
+      "loss": 0.5065,
+      "step": 408
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.927099942068421e-05,
+      "loss": 0.6068,
+      "step": 409
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9266270680892185e-05,
+      "loss": 0.586,
+      "step": 410
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9261527237903297e-05,
+      "loss": 0.5721,
+      "step": 411
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.925676909924418e-05,
+      "loss": 0.5171,
+      "step": 412
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9251996272464767e-05,
+      "loss": 0.5408,
+      "step": 413
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9247208765138325e-05,
+      "loss": 0.5806,
+      "step": 414
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9242406584861397e-05,
+      "loss": 0.6151,
+      "step": 415
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.923758973925381e-05,
+      "loss": 0.5776,
+      "step": 416
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.923275823595866e-05,
+      "loss": 0.5203,
+      "step": 417
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9227912082642313e-05,
+      "loss": 0.6244,
+      "step": 418
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9223051286994368e-05,
+      "loss": 0.6469,
+      "step": 419
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.921817585672766e-05,
+      "loss": 0.5401,
+      "step": 420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9213285799578255e-05,
+      "loss": 0.4647,
+      "step": 421
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9208381123305407e-05,
+      "loss": 0.5378,
+      "step": 422
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9203461835691596e-05,
+      "loss": 0.6897,
+      "step": 423
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9198527944542462e-05,
+      "loss": 0.5291,
+      "step": 424
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9193579457686824e-05,
+      "loss": 0.4973,
+      "step": 425
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.918861638297667e-05,
+      "loss": 0.7291,
+      "step": 426
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9183638728287118e-05,
+      "loss": 0.6306,
+      "step": 427
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.917864650151644e-05,
+      "loss": 0.5859,
+      "step": 428
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9173639710586015e-05,
+      "loss": 0.5858,
+      "step": 429
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9168618363440337e-05,
+      "loss": 0.6065,
+      "step": 430
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9163582468047003e-05,
+      "loss": 0.6478,
+      "step": 431
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9158532032396676e-05,
+      "loss": 0.5294,
+      "step": 432
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9153467064503116e-05,
+      "loss": 0.6138,
+      "step": 433
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9148387572403123e-05,
+      "loss": 0.7277,
+      "step": 434
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9143293564156555e-05,
+      "loss": 0.504,
+      "step": 435
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9138185047846284e-05,
+      "loss": 0.5224,
+      "step": 436
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9133062031578233e-05,
+      "loss": 0.6252,
+      "step": 437
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.91279245234813e-05,
+      "loss": 0.6277,
+      "step": 438
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9122772531707405e-05,
+      "loss": 0.6192,
+      "step": 439
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9117606064431436e-05,
+      "loss": 0.3334,
+      "step": 440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9112425129851247e-05,
+      "loss": 0.587,
+      "step": 441
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9107229736187666e-05,
+      "loss": 0.521,
+      "step": 442
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9102019891684438e-05,
+      "loss": 0.582,
+      "step": 443
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9096795604608258e-05,
+      "loss": 0.6046,
+      "step": 444
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.909155688324873e-05,
+      "loss": 0.5647,
+      "step": 445
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9086303735918358e-05,
+      "loss": 0.5645,
+      "step": 446
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.908103617095255e-05,
+      "loss": 0.7265,
+      "step": 447
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9075754196709574e-05,
+      "loss": 0.4044,
+      "step": 448
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9070457821570566e-05,
+      "loss": 0.4809,
+      "step": 449
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.906514705393953e-05,
+      "loss": 0.5613,
+      "step": 450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9059821902243277e-05,
+      "loss": 0.5482,
+      "step": 451
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.905448237493147e-05,
+      "loss": 0.6579,
+      "step": 452
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9049128480476563e-05,
+      "loss": 0.7165,
+      "step": 453
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9043760227373817e-05,
+      "loss": 0.5702,
+      "step": 454
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9038377624141276e-05,
+      "loss": 0.5408,
+      "step": 455
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.903298067931975e-05,
+      "loss": 0.5145,
+      "step": 456
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9027569401472806e-05,
+      "loss": 0.6121,
+      "step": 457
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.902214379918676e-05,
+      "loss": 0.538,
+      "step": 458
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9016703881070646e-05,
+      "loss": 0.5542,
+      "step": 459
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9011249655756224e-05,
+      "loss": 0.6698,
+      "step": 460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9005781131897953e-05,
+      "loss": 0.7971,
+      "step": 461
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9000298318172974e-05,
+      "loss": 0.5783,
+      "step": 462
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8994801223281113e-05,
+      "loss": 0.6863,
+      "step": 463
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8989289855944846e-05,
+      "loss": 0.4618,
+      "step": 464
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8983764224909302e-05,
+      "loss": 0.5024,
+      "step": 465
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8978224338942243e-05,
+      "loss": 0.6876,
+      "step": 466
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8972670206834048e-05,
+      "loss": 0.6186,
+      "step": 467
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8967101837397696e-05,
+      "loss": 0.6013,
+      "step": 468
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.896151923946877e-05,
+      "loss": 0.4508,
+      "step": 469
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8955922421905416e-05,
+      "loss": 0.6678,
+      "step": 470
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8950311393588357e-05,
+      "loss": 0.5649,
+      "step": 471
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8944686163420848e-05,
+      "loss": 0.4518,
+      "step": 472
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8939046740328692e-05,
+      "loss": 0.5052,
+      "step": 473
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.893339313326021e-05,
+      "loss": 0.5585,
+      "step": 474
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8927725351186223e-05,
+      "loss": 0.4745,
+      "step": 475
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8922043403100056e-05,
+      "loss": 0.5447,
+      "step": 476
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8916347298017493e-05,
+      "loss": 0.5858,
+      "step": 477
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8910637044976802e-05,
+      "loss": 0.5215,
+      "step": 478
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.890491265303869e-05,
+      "loss": 0.45,
+      "step": 479
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8899174131286298e-05,
+      "loss": 0.557,
+      "step": 480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.889342148882519e-05,
+      "loss": 0.7172,
+      "step": 481
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8887654734783338e-05,
+      "loss": 0.4896,
+      "step": 482
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8881873878311095e-05,
+      "loss": 0.4198,
+      "step": 483
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8876078928581207e-05,
+      "loss": 0.7311,
+      "step": 484
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.887026989478877e-05,
+      "loss": 0.6153,
+      "step": 485
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8864446786151232e-05,
+      "loss": 0.4316,
+      "step": 486
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.885860961190838e-05,
+      "loss": 0.5991,
+      "step": 487
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8852758381322305e-05,
+      "loss": 0.4792,
+      "step": 488
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8846893103677414e-05,
+      "loss": 0.6838,
+      "step": 489
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.88410137882804e-05,
+      "loss": 0.5466,
+      "step": 490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.883512044446023e-05,
+      "loss": 0.6053,
+      "step": 491
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8829213081568133e-05,
+      "loss": 0.5409,
+      "step": 492
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.882329170897757e-05,
+      "loss": 0.5552,
+      "step": 493
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8817356336084252e-05,
+      "loss": 0.6592,
+      "step": 494
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8811406972306092e-05,
+      "loss": 0.4795,
+      "step": 495
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8805443627083203e-05,
+      "loss": 0.4956,
+      "step": 496
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.879946630987789e-05,
+      "loss": 0.5265,
+      "step": 497
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.879347503017462e-05,
+      "loss": 0.4017,
+      "step": 498
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.878746979748002e-05,
+      "loss": 0.5318,
+      "step": 499
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.878145062132285e-05,
+      "loss": 0.5603,
+      "step": 500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.877541751125401e-05,
+      "loss": 0.4803,
+      "step": 501
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8769370476846496e-05,
+      "loss": 0.6391,
+      "step": 502
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8763309527695395e-05,
+      "loss": 0.5012,
+      "step": 503
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8757234673417892e-05,
+      "loss": 0.495,
+      "step": 504
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8751145923653214e-05,
+      "loss": 0.4807,
+      "step": 505
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8745043288062646e-05,
+      "loss": 0.4796,
+      "step": 506
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8738926776329517e-05,
+      "loss": 0.4857,
+      "step": 507
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8732796398159154e-05,
+      "loss": 0.6295,
+      "step": 508
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.87266521632789e-05,
+      "loss": 0.469,
+      "step": 509
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.872049408143808e-05,
+      "loss": 0.5097,
+      "step": 510
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8714322162407988e-05,
+      "loss": 0.5067,
+      "step": 511
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8708136415981885e-05,
+      "loss": 0.6171,
+      "step": 512
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8701936851974964e-05,
+      "loss": 0.5916,
+      "step": 513
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.869572348022434e-05,
+      "loss": 0.5447,
+      "step": 514
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.868949631058905e-05,
+      "loss": 0.661,
+      "step": 515
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8683255352950004e-05,
+      "loss": 0.544,
+      "step": 516
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.867700061721001e-05,
+      "loss": 0.6592,
+      "step": 517
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.867073211329373e-05,
+      "loss": 0.5091,
+      "step": 518
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8664449851147675e-05,
+      "loss": 0.4888,
+      "step": 519
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8658153840740182e-05,
+      "loss": 0.5722,
+      "step": 520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8651844092061406e-05,
+      "loss": 0.5019,
+      "step": 521
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.86455206151233e-05,
+      "loss": 0.4619,
+      "step": 522
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8639183419959606e-05,
+      "loss": 0.7055,
+      "step": 523
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8632832516625825e-05,
+      "loss": 0.5267,
+      "step": 524
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8626467915199216e-05,
+      "loss": 0.4979,
+      "step": 525
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8620089625778764e-05,
+      "loss": 0.5436,
+      "step": 526
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8613697658485188e-05,
+      "loss": 0.5754,
+      "step": 527
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8607292023460897e-05,
+      "loss": 0.5685,
+      "step": 528
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8600872730869995e-05,
+      "loss": 0.5778,
+      "step": 529
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.859443979089825e-05,
+      "loss": 0.5853,
+      "step": 530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.858799321375309e-05,
+      "loss": 0.6459,
+      "step": 531
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8581533009663587e-05,
+      "loss": 0.4425,
+      "step": 532
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8575059188880425e-05,
+      "loss": 0.5512,
+      "step": 533
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8568571761675893e-05,
+      "loss": 0.4594,
+      "step": 534
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8562070738343886e-05,
+      "loss": 0.4119,
+      "step": 535
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8555556129199855e-05,
+      "loss": 0.5311,
+      "step": 536
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8549027944580814e-05,
+      "loss": 0.6144,
+      "step": 537
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8542486194845326e-05,
+      "loss": 0.6419,
+      "step": 538
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8535930890373467e-05,
+      "loss": 0.5258,
+      "step": 539
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8529362041566822e-05,
+      "loss": 0.5021,
+      "step": 540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.852277965884848e-05,
+      "loss": 0.6229,
+      "step": 541
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8516183752662985e-05,
+      "loss": 0.6621,
+      "step": 542
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8509574333476355e-05,
+      "loss": 0.4192,
+      "step": 543
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8502951411776043e-05,
+      "loss": 0.5671,
+      "step": 544
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8496314998070932e-05,
+      "loss": 0.4873,
+      "step": 545
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8489665102891306e-05,
+      "loss": 0.4564,
+      "step": 546
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8483001736788843e-05,
+      "loss": 0.5759,
+      "step": 547
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8476324910336602e-05,
+      "loss": 0.7072,
+      "step": 548
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8469634634128993e-05,
+      "loss": 0.6284,
+      "step": 549
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.846293091878177e-05,
+      "loss": 0.4938,
+      "step": 550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8456213774932012e-05,
+      "loss": 0.5306,
+      "step": 551
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8449483213238105e-05,
+      "loss": 0.7414,
+      "step": 552
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8442739244379724e-05,
+      "loss": 0.6226,
+      "step": 553
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.843598187905782e-05,
+      "loss": 0.6228,
+      "step": 554
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8429211127994596e-05,
+      "loss": 0.6275,
+      "step": 555
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8422427001933507e-05,
+      "loss": 0.4483,
+      "step": 556
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8415629511639223e-05,
+      "loss": 0.6363,
+      "step": 557
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8408818667897607e-05,
+      "loss": 0.5551,
+      "step": 558
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.840199448151573e-05,
+      "loss": 0.6698,
+      "step": 559
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8395156963321832e-05,
+      "loss": 0.4839,
+      "step": 560
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8388306124165294e-05,
+      "loss": 0.5133,
+      "step": 561
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8381441974916642e-05,
+      "loss": 0.4715,
+      "step": 562
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.837456452646753e-05,
+      "loss": 0.5085,
+      "step": 563
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.83676737897307e-05,
+      "loss": 0.6097,
+      "step": 564
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8360769775639983e-05,
+      "loss": 0.6361,
+      "step": 565
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8353852495150286e-05,
+      "loss": 0.5743,
+      "step": 566
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8346921959237555e-05,
+      "loss": 0.5667,
+      "step": 567
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.833997817889878e-05,
+      "loss": 0.5593,
+      "step": 568
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.833302116515196e-05,
+      "loss": 0.4797,
+      "step": 569
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8326050929036084e-05,
+      "loss": 0.3929,
+      "step": 570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.831906748161114e-05,
+      "loss": 0.4746,
+      "step": 571
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.831207083395807e-05,
+      "loss": 0.4555,
+      "step": 572
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8305060997178755e-05,
+      "loss": 0.6804,
+      "step": 573
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8298037982396013e-05,
+      "loss": 0.423,
+      "step": 574
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8291001800753565e-05,
+      "loss": 0.5597,
+      "step": 575
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8283952463416034e-05,
+      "loss": 0.5374,
+      "step": 576
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.827688998156891e-05,
+      "loss": 0.4442,
+      "step": 577
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8269814366418536e-05,
+      "loss": 0.4976,
+      "step": 578
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8262725629192105e-05,
+      "loss": 0.4193,
+      "step": 579
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8255623781137624e-05,
+      "loss": 0.5439,
+      "step": 580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.824850883352391e-05,
+      "loss": 0.4198,
+      "step": 581
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8241380797640555e-05,
+      "loss": 0.6482,
+      "step": 582
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8234239684797934e-05,
+      "loss": 0.6121,
+      "step": 583
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8227085506327157e-05,
+      "loss": 0.3949,
+      "step": 584
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8219918273580072e-05,
+      "loss": 0.5791,
+      "step": 585
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8212737997929238e-05,
+      "loss": 0.8548,
+      "step": 586
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.820554469076792e-05,
+      "loss": 0.5771,
+      "step": 587
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8198338363510046e-05,
+      "loss": 0.5826,
+      "step": 588
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.819111902759022e-05,
+      "loss": 0.538,
+      "step": 589
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.818388669446366e-05,
+      "loss": 0.495,
+      "step": 590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8176641375606244e-05,
+      "loss": 0.5056,
+      "step": 591
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.816938308251443e-05,
+      "loss": 0.4531,
+      "step": 592
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8162111826705265e-05,
+      "loss": 0.5761,
+      "step": 593
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.815482761971637e-05,
+      "loss": 0.5132,
+      "step": 594
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8147530473105917e-05,
+      "loss": 0.6751,
+      "step": 595
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8140220398452607e-05,
+      "loss": 0.4405,
+      "step": 596
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8132897407355657e-05,
+      "loss": 0.4562,
+      "step": 597
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.812556151143477e-05,
+      "loss": 0.5395,
+      "step": 598
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8118212722330143e-05,
+      "loss": 0.6424,
+      "step": 599
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.811085105170241e-05,
+      "loss": 0.4635,
+      "step": 600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8103476511232666e-05,
+      "loss": 0.6133,
+      "step": 601
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.809608911262241e-05,
+      "loss": 0.6785,
+      "step": 602
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8088688867593553e-05,
+      "loss": 0.5665,
+      "step": 603
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8081275787888385e-05,
+      "loss": 0.4655,
+      "step": 604
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.807384988526957e-05,
+      "loss": 0.4033,
+      "step": 605
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8066411171520107e-05,
+      "loss": 0.5162,
+      "step": 606
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8058959658443327e-05,
+      "loss": 0.5424,
+      "step": 607
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8051495357862877e-05,
+      "loss": 0.5145,
+      "step": 608
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.804401828162268e-05,
+      "loss": 0.4072,
+      "step": 609
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8036528441586955e-05,
+      "loss": 0.7015,
+      "step": 610
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8029025849640143e-05,
+      "loss": 0.554,
+      "step": 611
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8021510517686945e-05,
+      "loss": 0.6613,
+      "step": 612
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.801398245765226e-05,
+      "loss": 0.5778,
+      "step": 613
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8006441681481194e-05,
+      "loss": 0.4456,
+      "step": 614
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7998888201139027e-05,
+      "loss": 0.5531,
+      "step": 615
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7991322028611196e-05,
+      "loss": 0.4384,
+      "step": 616
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7983743175903274e-05,
+      "loss": 0.5251,
+      "step": 617
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7976151655040965e-05,
+      "loss": 0.5648,
+      "step": 618
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.796854747807006e-05,
+      "loss": 0.4709,
+      "step": 619
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.796093065705644e-05,
+      "loss": 0.5251,
+      "step": 620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7953301204086052e-05,
+      "loss": 0.4508,
+      "step": 621
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.794565913126488e-05,
+      "loss": 0.5798,
+      "step": 622
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7938004450718934e-05,
+      "loss": 0.5568,
+      "step": 623
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7930337174594234e-05,
+      "loss": 0.5074,
+      "step": 624
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.792265731505678e-05,
+      "loss": 0.5943,
+      "step": 625
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7914964884292543e-05,
+      "loss": 0.4738,
+      "step": 626
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7907259894507435e-05,
+      "loss": 0.4776,
+      "step": 627
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7899542357927303e-05,
+      "loss": 0.4672,
+      "step": 628
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7891812286797902e-05,
+      "loss": 0.6533,
+      "step": 629
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7884069693384866e-05,
+      "loss": 0.5309,
+      "step": 630
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7876314589973717e-05,
+      "loss": 0.5415,
+      "step": 631
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.786854698886981e-05,
+      "loss": 0.4401,
+      "step": 632
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7860766902398338e-05,
+      "loss": 0.5742,
+      "step": 633
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7852974342904303e-05,
+      "loss": 0.4586,
+      "step": 634
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7845169322752504e-05,
+      "loss": 0.4853,
+      "step": 635
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.783735185432751e-05,
+      "loss": 0.5721,
+      "step": 636
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.782952195003363e-05,
+      "loss": 0.6234,
+      "step": 637
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7821679622294925e-05,
+      "loss": 0.5585,
+      "step": 638
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7813824883555152e-05,
+      "loss": 0.5604,
+      "step": 639
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7805957746277776e-05,
+      "loss": 0.4236,
+      "step": 640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.779807822294592e-05,
+      "loss": 0.5173,
+      "step": 641
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7790186326062375e-05,
+      "loss": 0.6531,
+      "step": 642
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7782282068149555e-05,
+      "loss": 0.4605,
+      "step": 643
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.777436546174949e-05,
+      "loss": 0.431,
+      "step": 644
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.776643651942381e-05,
+      "loss": 0.5807,
+      "step": 645
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7758495253753713e-05,
+      "loss": 0.4999,
+      "step": 646
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7750541677339953e-05,
+      "loss": 0.5812,
+      "step": 647
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.774257580280281e-05,
+      "loss": 0.3536,
+      "step": 648
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7734597642782102e-05,
+      "loss": 0.4341,
+      "step": 649
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.772660720993711e-05,
+      "loss": 0.59,
+      "step": 650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7718604516946604e-05,
+      "loss": 0.6065,
+      "step": 651
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.771058957650881e-05,
+      "loss": 0.6338,
+      "step": 652
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7702562401341386e-05,
+      "loss": 0.5928,
+      "step": 653
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7694523004181406e-05,
+      "loss": 0.5554,
+      "step": 654
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7686471397785322e-05,
+      "loss": 0.4562,
+      "step": 655
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7678407594928975e-05,
+      "loss": 0.5701,
+      "step": 656
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7670331608407558e-05,
+      "loss": 0.4164,
+      "step": 657
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7662243451035584e-05,
+      "loss": 0.4733,
+      "step": 658
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7654143135646894e-05,
+      "loss": 0.6205,
+      "step": 659
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.764603067509461e-05,
+      "loss": 0.6272,
+      "step": 660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.763790608225112e-05,
+      "loss": 0.4633,
+      "step": 661
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.762976937000808e-05,
+      "loss": 0.5721,
+      "step": 662
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7621620551276366e-05,
+      "loss": 0.591,
+      "step": 663
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7613459638986064e-05,
+      "loss": 0.4893,
+      "step": 664
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7605286646086446e-05,
+      "loss": 0.6291,
+      "step": 665
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7597101585545962e-05,
+      "loss": 0.6488,
+      "step": 666
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7588904470352206e-05,
+      "loss": 0.5109,
+      "step": 667
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7580695313511897e-05,
+      "loss": 0.3789,
+      "step": 668
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7572474128050864e-05,
+      "loss": 0.5143,
+      "step": 669
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.756424092701402e-05,
+      "loss": 0.5304,
+      "step": 670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7555995723465344e-05,
+      "loss": 0.4122,
+      "step": 671
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7547738530487867e-05,
+      "loss": 0.5394,
+      "step": 672
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.753946936118363e-05,
+      "loss": 0.4984,
+      "step": 673
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.753118822867369e-05,
+      "loss": 0.6238,
+      "step": 674
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.752289514609809e-05,
+      "loss": 0.5703,
+      "step": 675
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.751459012661581e-05,
+      "loss": 0.5751,
+      "step": 676
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.75062731834048e-05,
+      "loss": 0.4273,
+      "step": 677
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.749794432966191e-05,
+      "loss": 0.5555,
+      "step": 678
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.74896035786029e-05,
+      "loss": 0.5496,
+      "step": 679
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7481250943462408e-05,
+      "loss": 0.5731,
+      "step": 680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7472886437493922e-05,
+      "loss": 0.4722,
+      "step": 681
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7464510073969767e-05,
+      "loss": 0.5248,
+      "step": 682
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7456121866181087e-05,
+      "loss": 0.5678,
+      "step": 683
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.744772182743782e-05,
+      "loss": 0.4595,
+      "step": 684
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7439309971068673e-05,
+      "loss": 0.4879,
+      "step": 685
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7430886310421103e-05,
+      "loss": 0.5379,
+      "step": 686
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.742245085886131e-05,
+      "loss": 0.5538,
+      "step": 687
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7414003629774185e-05,
+      "loss": 0.5077,
+      "step": 688
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.740554463656332e-05,
+      "loss": 0.705,
+      "step": 689
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.739707389265097e-05,
+      "loss": 0.6453,
+      "step": 690
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7388591411478025e-05,
+      "loss": 0.4266,
+      "step": 691
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7380097206504025e-05,
+      "loss": 0.5832,
+      "step": 692
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7371591291207085e-05,
+      "loss": 0.5966,
+      "step": 693
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7363073679083918e-05,
+      "loss": 0.4466,
+      "step": 694
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.735454438364979e-05,
+      "loss": 0.5533,
+      "step": 695
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7346003418438502e-05,
+      "loss": 0.6313,
+      "step": 696
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7337450797002385e-05,
+      "loss": 0.5605,
+      "step": 697
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7328886532912254e-05,
+      "loss": 0.5726,
+      "step": 698
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7320310639757397e-05,
+      "loss": 0.4356,
+      "step": 699
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7311723131145564e-05,
+      "loss": 0.4874,
+      "step": 700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7303124020702935e-05,
+      "loss": 0.5377,
+      "step": 701
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7294513322074084e-05,
+      "loss": 0.592,
+      "step": 702
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.728589104892199e-05,
+      "loss": 0.6307,
+      "step": 703
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7277257214927987e-05,
+      "loss": 0.5054,
+      "step": 704
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7268611833791758e-05,
+      "loss": 0.4068,
+      "step": 705
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.725995491923131e-05,
+      "loss": 0.6134,
+      "step": 706
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7251286484982944e-05,
+      "loss": 0.6112,
+      "step": 707
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7242606544801247e-05,
+      "loss": 0.5295,
+      "step": 708
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7233915112459055e-05,
+      "loss": 0.5811,
+      "step": 709
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.722521220174745e-05,
+      "loss": 0.4005,
+      "step": 710
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7216497826475717e-05,
+      "loss": 0.5465,
+      "step": 711
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7207772000471338e-05,
+      "loss": 0.5566,
+      "step": 712
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7199034737579962e-05,
+      "loss": 0.5492,
+      "step": 713
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.719028605166538e-05,
+      "loss": 0.561,
+      "step": 714
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7181525956609527e-05,
+      "loss": 0.4006,
+      "step": 715
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7172754466312417e-05,
+      "loss": 0.5989,
+      "step": 716
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7163971594692167e-05,
+      "loss": 0.5182,
+      "step": 717
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7155177355684935e-05,
+      "loss": 0.5954,
+      "step": 718
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7146371763244932e-05,
+      "loss": 0.6256,
+      "step": 719
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.713755483134437e-05,
+      "loss": 0.549,
+      "step": 720
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.712872657397346e-05,
+      "loss": 0.5166,
+      "step": 721
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7119887005140392e-05,
+      "loss": 0.4976,
+      "step": 722
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.711103613887129e-05,
+      "loss": 0.5915,
+      "step": 723
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7102173989210212e-05,
+      "loss": 0.4497,
+      "step": 724
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.709330057021912e-05,
+      "loss": 0.5097,
+      "step": 725
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.708441589597785e-05,
+      "loss": 0.4605,
+      "step": 726
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7075519980584115e-05,
+      "loss": 0.4205,
+      "step": 727
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.706661283815344e-05,
+      "loss": 0.7057,
+      "step": 728
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7057694482819186e-05,
+      "loss": 0.5806,
+      "step": 729
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.704876492873249e-05,
+      "loss": 0.5076,
+      "step": 730
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7039824190062272e-05,
+      "loss": 0.4887,
+      "step": 731
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.703087228099519e-05,
+      "loss": 0.4517,
+      "step": 732
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.702190921573563e-05,
+      "loss": 0.5686,
+      "step": 733
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7012935008505685e-05,
+      "loss": 0.4267,
+      "step": 734
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7003949673545115e-05,
+      "loss": 0.5769,
+      "step": 735
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6994953225111346e-05,
+      "loss": 0.5335,
+      "step": 736
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6985945677479434e-05,
+      "loss": 0.6602,
+      "step": 737
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6976927044942054e-05,
+      "loss": 0.579,
+      "step": 738
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.696789734180946e-05,
+      "loss": 0.5008,
+      "step": 739
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6958856582409474e-05,
+      "loss": 0.5771,
+      "step": 740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.694980478108747e-05,
+      "loss": 0.5657,
+      "step": 741
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6940741952206342e-05,
+      "loss": 0.623,
+      "step": 742
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6931668110146465e-05,
+      "loss": 0.5751,
+      "step": 743
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6922583269305706e-05,
+      "loss": 0.6506,
+      "step": 744
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6913487444099382e-05,
+      "loss": 0.64,
+      "step": 745
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.690438064896023e-05,
+      "loss": 0.6004,
+      "step": 746
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.689526289833841e-05,
+      "loss": 0.6683,
+      "step": 747
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6886134206701444e-05,
+      "loss": 0.5072,
+      "step": 748
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6876994588534234e-05,
+      "loss": 0.4651,
+      "step": 749
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.686784405833901e-05,
+      "loss": 0.4564,
+      "step": 750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6858682630635317e-05,
+      "loss": 0.738,
+      "step": 751
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6849510319959997e-05,
+      "loss": 0.5601,
+      "step": 752
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.684032714086715e-05,
+      "loss": 0.4036,
+      "step": 753
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.683113310792813e-05,
+      "loss": 0.5975,
+      "step": 754
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6821928235731518e-05,
+      "loss": 0.5097,
+      "step": 755
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6812712538883075e-05,
+      "loss": 0.5343,
+      "step": 756
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6803486032005758e-05,
+      "loss": 0.4901,
+      "step": 757
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.679424872973967e-05,
+      "loss": 0.4307,
+      "step": 758
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6785000646742042e-05,
+      "loss": 0.489,
+      "step": 759
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6775741797687205e-05,
+      "loss": 0.5298,
+      "step": 760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6766472197266583e-05,
+      "loss": 0.4619,
+      "step": 761
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.675719186018866e-05,
+      "loss": 0.4893,
+      "step": 762
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.674790080117895e-05,
+      "loss": 0.4172,
+      "step": 763
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6738599034979984e-05,
+      "loss": 0.4681,
+      "step": 764
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6729286576351278e-05,
+      "loss": 0.4714,
+      "step": 765
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.671996344006932e-05,
+      "loss": 0.5558,
+      "step": 766
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.671062964092754e-05,
+      "loss": 0.5791,
+      "step": 767
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6701285193736277e-05,
+      "loss": 0.5731,
+      "step": 768
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6691930113322784e-05,
+      "loss": 0.5212,
+      "step": 769
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.668256441453117e-05,
+      "loss": 0.4011,
+      "step": 770
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6673188112222394e-05,
+      "loss": 0.5109,
+      "step": 771
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.666380122127426e-05,
+      "loss": 0.5214,
+      "step": 772
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6654403756581344e-05,
+      "loss": 0.6231,
+      "step": 773
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6644995733055022e-05,
+      "loss": 0.5574,
+      "step": 774
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6635577165623413e-05,
+      "loss": 0.5235,
+      "step": 775
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6626148069231372e-05,
+      "loss": 0.4963,
+      "step": 776
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6616708458840456e-05,
+      "loss": 0.5322,
+      "step": 777
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.660725834942891e-05,
+      "loss": 0.6456,
+      "step": 778
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6597797755991634e-05,
+      "loss": 0.5383,
+      "step": 779
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6588326693540165e-05,
+      "loss": 0.5368,
+      "step": 780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.657884517710265e-05,
+      "loss": 0.5253,
+      "step": 781
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6569353221723833e-05,
+      "loss": 0.5335,
+      "step": 782
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.655985084246501e-05,
+      "loss": 0.542,
+      "step": 783
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6550338054404022e-05,
+      "loss": 0.4661,
+      "step": 784
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6540814872635224e-05,
+      "loss": 0.563,
+      "step": 785
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.653128131226947e-05,
+      "loss": 0.5332,
+      "step": 786
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.652173738843407e-05,
+      "loss": 0.5527,
+      "step": 787
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6512183116272786e-05,
+      "loss": 0.465,
+      "step": 788
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6502618510945808e-05,
+      "loss": 0.5448,
+      "step": 789
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6493043587629704e-05,
+      "loss": 0.6032,
+      "step": 790
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.648345836151743e-05,
+      "loss": 0.5896,
+      "step": 791
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.647386284781828e-05,
+      "loss": 0.5394,
+      "step": 792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6464257061757873e-05,
+      "loss": 0.626,
+      "step": 793
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6454641018578135e-05,
+      "loss": 0.561,
+      "step": 794
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.644501473353726e-05,
+      "loss": 0.5696,
+      "step": 795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.643537822190969e-05,
+      "loss": 0.4591,
+      "step": 796
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6425731498986112e-05,
+      "loss": 0.6469,
+      "step": 797
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6416074580073393e-05,
+      "loss": 0.6272,
+      "step": 798
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.640640748049459e-05,
+      "loss": 0.3535,
+      "step": 799
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6396730215588913e-05,
+      "loss": 0.4854,
+      "step": 800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6387042800711706e-05,
+      "loss": 0.3388,
+      "step": 801
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6377345251234403e-05,
+      "loss": 0.5252,
+      "step": 802
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6367637582544545e-05,
+      "loss": 0.4378,
+      "step": 803
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.63579198100457e-05,
+      "loss": 0.4599,
+      "step": 804
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.634819194915749e-05,
+      "loss": 0.5409,
+      "step": 805
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6338454015315535e-05,
+      "loss": 0.4836,
+      "step": 806
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6328706023971444e-05,
+      "loss": 0.5209,
+      "step": 807
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6318947990592776e-05,
+      "loss": 0.4886,
+      "step": 808
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.630917993066303e-05,
+      "loss": 0.3754,
+      "step": 809
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6299401859681617e-05,
+      "loss": 0.5332,
+      "step": 810
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6289613793163827e-05,
+      "loss": 0.4685,
+      "step": 811
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6279815746640814e-05,
+      "loss": 0.5962,
+      "step": 812
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.627000773565957e-05,
+      "loss": 0.57,
+      "step": 813
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6260189775782888e-05,
+      "loss": 0.5643,
+      "step": 814
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6250361882589364e-05,
+      "loss": 0.4716,
+      "step": 815
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6240524071673337e-05,
+      "loss": 0.4474,
+      "step": 816
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6230676358644895e-05,
+      "loss": 0.4918,
+      "step": 817
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.622081875912983e-05,
+      "loss": 0.5463,
+      "step": 818
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.621095128876964e-05,
+      "loss": 0.5585,
+      "step": 819
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6201073963221455e-05,
+      "loss": 0.5705,
+      "step": 820
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.619118679815807e-05,
+      "loss": 0.4306,
+      "step": 821
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6181289809267875e-05,
+      "loss": 0.5045,
+      "step": 822
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6171383012254862e-05,
+      "loss": 0.4123,
+      "step": 823
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6161466422838578e-05,
+      "loss": 0.5769,
+      "step": 824
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6151540056754103e-05,
+      "loss": 0.5839,
+      "step": 825
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6141603929752043e-05,
+      "loss": 0.5781,
+      "step": 826
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6131658057598483e-05,
+      "loss": 0.4735,
+      "step": 827
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.612170245607497e-05,
+      "loss": 0.6236,
+      "step": 828
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6111737140978495e-05,
+      "loss": 0.3984,
+      "step": 829
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.610176212812146e-05,
+      "loss": 0.6263,
+      "step": 830
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6091777433331652e-05,
+      "loss": 0.4873,
+      "step": 831
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6081783072452223e-05,
+      "loss": 0.5363,
+      "step": 832
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6071779061341665e-05,
+      "loss": 0.5563,
+      "step": 833
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.606176541587378e-05,
+      "loss": 0.3984,
+      "step": 834
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6051742151937655e-05,
+      "loss": 0.5054,
+      "step": 835
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6041709285437635e-05,
+      "loss": 0.539,
+      "step": 836
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6031666832293322e-05,
+      "loss": 0.6398,
+      "step": 837
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6021614808439505e-05,
+      "loss": 0.5973,
+      "step": 838
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6011553229826174e-05,
+      "loss": 0.4819,
+      "step": 839
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.600148211241847e-05,
+      "loss": 0.4556,
+      "step": 840
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5991401472196687e-05,
+      "loss": 0.4838,
+      "step": 841
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5981311325156205e-05,
+      "loss": 0.6422,
+      "step": 842
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5971211687307508e-05,
+      "loss": 0.5264,
+      "step": 843
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.596110257467613e-05,
+      "loss": 0.5783,
+      "step": 844
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.595098400330265e-05,
+      "loss": 0.6364,
+      "step": 845
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5940855989242636e-05,
+      "loss": 0.5087,
+      "step": 846
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5930718548566657e-05,
+      "loss": 0.5616,
+      "step": 847
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5920571697360235e-05,
+      "loss": 0.5191,
+      "step": 848
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.591041545172382e-05,
+      "loss": 0.5307,
+      "step": 849
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.590024982777277e-05,
+      "loss": 0.5379,
+      "step": 850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5890074841637328e-05,
+      "loss": 0.565,
+      "step": 851
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5879890509462582e-05,
+      "loss": 0.4052,
+      "step": 852
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5869696847408468e-05,
+      "loss": 0.4832,
+      "step": 853
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.585949387164971e-05,
+      "loss": 0.5109,
+      "step": 854
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5849281598375816e-05,
+      "loss": 0.4731,
+      "step": 855
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5839060043791047e-05,
+      "loss": 0.5507,
+      "step": 856
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5828829224114387e-05,
+      "loss": 0.6051,
+      "step": 857
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.581858915557953e-05,
+      "loss": 0.5241,
+      "step": 858
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.580833985443484e-05,
+      "loss": 0.583,
+      "step": 859
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5798081336943325e-05,
+      "loss": 0.5509,
+      "step": 860
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.578781361938263e-05,
+      "loss": 0.585,
+      "step": 861
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5777536718044992e-05,
+      "loss": 0.5262,
+      "step": 862
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.576725064923722e-05,
+      "loss": 0.4947,
+      "step": 863
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5756955429280657e-05,
+      "loss": 0.4898,
+      "step": 864
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5746651074511197e-05,
+      "loss": 0.5461,
+      "step": 865
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5736337601279194e-05,
+      "loss": 0.455,
+      "step": 866
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.57260150259495e-05,
+      "loss": 0.2914,
+      "step": 867
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5715683364901387e-05,
+      "loss": 0.2908,
+      "step": 868
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5705342634528558e-05,
+      "loss": 0.2848,
+      "step": 869
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5694992851239102e-05,
+      "loss": 0.2967,
+      "step": 870
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.568463403145547e-05,
+      "loss": 0.2897,
+      "step": 871
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.567426619161446e-05,
+      "loss": 0.3155,
+      "step": 872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5663889348167168e-05,
+      "loss": 0.2266,
+      "step": 873
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5653503517578995e-05,
+      "loss": 0.3028,
+      "step": 874
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5643108716329588e-05,
+      "loss": 0.285,
+      "step": 875
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.563270496091283e-05,
+      "loss": 0.2885,
+      "step": 876
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5622292267836815e-05,
+      "loss": 0.2261,
+      "step": 877
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5611870653623826e-05,
+      "loss": 0.2662,
+      "step": 878
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5601440134810286e-05,
+      "loss": 0.3377,
+      "step": 879
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5591000727946756e-05,
+      "loss": 0.2319,
+      "step": 880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5580552449597902e-05,
+      "loss": 0.2475,
+      "step": 881
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5570095316342464e-05,
+      "loss": 0.3172,
+      "step": 882
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5559629344773233e-05,
+      "loss": 0.2893,
+      "step": 883
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.554915455149702e-05,
+      "loss": 0.2457,
+      "step": 884
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5538670953134638e-05,
+      "loss": 0.3201,
+      "step": 885
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5528178566320877e-05,
+      "loss": 0.276,
+      "step": 886
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.551767740770446e-05,
+      "loss": 0.3369,
+      "step": 887
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5507167493948035e-05,
+      "loss": 0.2995,
+      "step": 888
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5496648841728147e-05,
+      "loss": 0.2996,
+      "step": 889
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5486121467735197e-05,
+      "loss": 0.2489,
+      "step": 890
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5475585388673428e-05,
+      "loss": 0.3433,
+      "step": 891
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5465040621260897e-05,
+      "loss": 0.2337,
+      "step": 892
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5454487182229456e-05,
+      "loss": 0.2585,
+      "step": 893
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5443925088324697e-05,
+      "loss": 0.2373,
+      "step": 894
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5433354356305966e-05,
+      "loss": 0.2436,
+      "step": 895
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5422775002946298e-05,
+      "loss": 0.2297,
+      "step": 896
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5412187045032424e-05,
+      "loss": 0.2498,
+      "step": 897
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5401590499364707e-05,
+      "loss": 0.3336,
+      "step": 898
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.539098538275716e-05,
+      "loss": 0.2622,
+      "step": 899
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.538037171203738e-05,
+      "loss": 0.2942,
+      "step": 900
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.536974950404655e-05,
+      "loss": 0.2928,
+      "step": 901
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5359118775639382e-05,
+      "loss": 0.2846,
+      "step": 902
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5348479543684123e-05,
+      "loss": 0.2977,
+      "step": 903
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5337831825062507e-05,
+      "loss": 0.2686,
+      "step": 904
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5327175636669733e-05,
+      "loss": 0.3893,
+      "step": 905
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5316510995414443e-05,
+      "loss": 0.2553,
+      "step": 906
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.530583791821869e-05,
+      "loss": 0.265,
+      "step": 907
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5295156422017908e-05,
+      "loss": 0.2188,
+      "step": 908
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.52844665237609e-05,
+      "loss": 0.2336,
+      "step": 909
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5273768240409792e-05,
+      "loss": 0.2646,
+      "step": 910
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.526306158894002e-05,
+      "loss": 0.2196,
+      "step": 911
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5252346586340293e-05,
+      "loss": 0.3654,
+      "step": 912
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5241623249612572e-05,
+      "loss": 0.2333,
+      "step": 913
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.523089159577205e-05,
+      "loss": 0.2591,
+      "step": 914
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.522015164184711e-05,
+      "loss": 0.3351,
+      "step": 915
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5209403404879305e-05,
+      "loss": 0.2926,
+      "step": 916
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5198646901923333e-05,
+      "loss": 0.2713,
+      "step": 917
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5187882150047006e-05,
+      "loss": 0.2174,
+      "step": 918
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5177109166331232e-05,
+      "loss": 0.2701,
+      "step": 919
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.516632796786997e-05,
+      "loss": 0.2095,
+      "step": 920
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.515553857177022e-05,
+      "loss": 0.2876,
+      "step": 921
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5144740995151988e-05,
+      "loss": 0.2867,
+      "step": 922
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5133935255148264e-05,
+      "loss": 0.2775,
+      "step": 923
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5123121368904983e-05,
+      "loss": 0.2746,
+      "step": 924
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5112299353581019e-05,
+      "loss": 0.2248,
+      "step": 925
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5101469226348132e-05,
+      "loss": 0.2498,
+      "step": 926
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5090631004390957e-05,
+      "loss": 0.2554,
+      "step": 927
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5079784704906975e-05,
+      "loss": 0.3206,
+      "step": 928
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5068930345106486e-05,
+      "loss": 0.2835,
+      "step": 929
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5058067942212578e-05,
+      "loss": 0.3275,
+      "step": 930
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5047197513461095e-05,
+      "loss": 0.2742,
+      "step": 931
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5036319076100627e-05,
+      "loss": 0.2296,
+      "step": 932
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5025432647392468e-05,
+      "loss": 0.2227,
+      "step": 933
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5014538244610586e-05,
+      "loss": 0.3274,
+      "step": 934
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5003635885041608e-05,
+      "loss": 0.2878,
+      "step": 935
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4992725585984784e-05,
+      "loss": 0.2875,
+      "step": 936
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.498180736475196e-05,
+      "loss": 0.2525,
+      "step": 937
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4970881238667561e-05,
+      "loss": 0.273,
+      "step": 938
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4959947225068546e-05,
+      "loss": 0.3188,
+      "step": 939
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.494900534130439e-05,
+      "loss": 0.3076,
+      "step": 940
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.493805560473706e-05,
+      "loss": 0.2729,
+      "step": 941
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4927098032740984e-05,
+      "loss": 0.2559,
+      "step": 942
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.491613264270302e-05,
+      "loss": 0.2006,
+      "step": 943
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4905159452022431e-05,
+      "loss": 0.3197,
+      "step": 944
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4894178478110856e-05,
+      "loss": 0.2294,
+      "step": 945
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.488318973839229e-05,
+      "loss": 0.2341,
+      "step": 946
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.487219325030305e-05,
+      "loss": 0.3341,
+      "step": 947
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4861189031291732e-05,
+      "loss": 0.2208,
+      "step": 948
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4850177098819223e-05,
+      "loss": 0.2773,
+      "step": 949
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4839157470358638e-05,
+      "loss": 0.2415,
+      "step": 950
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4828130163395295e-05,
+      "loss": 0.2375,
+      "step": 951
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4817095195426712e-05,
+      "loss": 0.3083,
+      "step": 952
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.480605258396255e-05,
+      "loss": 0.2296,
+      "step": 953
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4795002346524608e-05,
+      "loss": 0.2614,
+      "step": 954
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4783944500646777e-05,
+      "loss": 0.2283,
+      "step": 955
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4772879063875024e-05,
+      "loss": 0.3157,
+      "step": 956
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4761806053767364e-05,
+      "loss": 0.3208,
+      "step": 957
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4750725487893826e-05,
+      "loss": 0.2712,
+      "step": 958
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4739637383836429e-05,
+      "loss": 0.3188,
+      "step": 959
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.472854175918915e-05,
+      "loss": 0.2696,
+      "step": 960
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4717438631557908e-05,
+      "loss": 0.2906,
+      "step": 961
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4706328018560511e-05,
+      "loss": 0.2593,
+      "step": 962
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4695209937826662e-05,
+      "loss": 0.2748,
+      "step": 963
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4684084406997903e-05,
+      "loss": 0.2325,
+      "step": 964
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.46729514437276e-05,
+      "loss": 0.2579,
+      "step": 965
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4661811065680916e-05,
+      "loss": 0.2581,
+      "step": 966
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4650663290534775e-05,
+      "loss": 0.2599,
+      "step": 967
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4639508135977834e-05,
+      "loss": 0.2291,
+      "step": 968
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4628345619710468e-05,
+      "loss": 0.2226,
+      "step": 969
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4617175759444732e-05,
+      "loss": 0.2997,
+      "step": 970
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4605998572904332e-05,
+      "loss": 0.3604,
+      "step": 971
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4594814077824592e-05,
+      "loss": 0.2422,
+      "step": 972
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4583622291952444e-05,
+      "loss": 0.2613,
+      "step": 973
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4572423233046386e-05,
+      "loss": 0.2064,
+      "step": 974
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4561216918876453e-05,
+      "loss": 0.2514,
+      "step": 975
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.455000336722419e-05,
+      "loss": 0.2351,
+      "step": 976
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4538782595882635e-05,
+      "loss": 0.2754,
+      "step": 977
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4527554622656273e-05,
+      "loss": 0.2342,
+      "step": 978
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.451631946536102e-05,
+      "loss": 0.2783,
+      "step": 979
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4505077141824191e-05,
+      "loss": 0.2274,
+      "step": 980
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4493827669884479e-05,
+      "loss": 0.2572,
+      "step": 981
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4482571067391905e-05,
+      "loss": 0.2802,
+      "step": 982
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4471307352207815e-05,
+      "loss": 0.2609,
+      "step": 983
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4460036542204838e-05,
+      "loss": 0.2602,
+      "step": 984
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4448758655266866e-05,
+      "loss": 0.2465,
+      "step": 985
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4437473709289006e-05,
+      "loss": 0.2229,
+      "step": 986
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4426181722177581e-05,
+      "loss": 0.3018,
+      "step": 987
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.441488271185008e-05,
+      "loss": 0.2286,
+      "step": 988
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4403576696235138e-05,
+      "loss": 0.3196,
+      "step": 989
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4392263693272503e-05,
+      "loss": 0.3198,
+      "step": 990
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4380943720913012e-05,
+      "loss": 0.2806,
+      "step": 991
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.436961679711856e-05,
+      "loss": 0.2092,
+      "step": 992
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.435828293986207e-05,
+      "loss": 0.2707,
+      "step": 993
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4346942167127469e-05,
+      "loss": 0.2857,
+      "step": 994
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4335594496909663e-05,
+      "loss": 0.294,
+      "step": 995
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.432423994721449e-05,
+      "loss": 0.2825,
+      "step": 996
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4312878536058712e-05,
+      "loss": 0.2756,
+      "step": 997
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.430151028146998e-05,
+      "loss": 0.3352,
+      "step": 998
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4290135201486796e-05,
+      "loss": 0.3088,
+      "step": 999
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.42787533141585e-05,
+      "loss": 0.2976,
+      "step": 1000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.426736463754523e-05,
+      "loss": 0.2443,
+      "step": 1001
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4255969189717895e-05,
+      "loss": 0.2972,
+      "step": 1002
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4244566988758152e-05,
+      "loss": 0.3158,
+      "step": 1003
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4233158052758372e-05,
+      "loss": 0.2928,
+      "step": 1004
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4221742399821614e-05,
+      "loss": 0.2852,
+      "step": 1005
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.421032004806159e-05,
+      "loss": 0.2272,
+      "step": 1006
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4198891015602648e-05,
+      "loss": 0.2863,
+      "step": 1007
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.418745532057973e-05,
+      "loss": 0.2703,
+      "step": 1008
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.417601298113836e-05,
+      "loss": 0.284,
+      "step": 1009
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4164564015434592e-05,
+      "loss": 0.2469,
+      "step": 1010
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4153108441635002e-05,
+      "loss": 0.2754,
+      "step": 1011
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.414164627791665e-05,
+      "loss": 0.3665,
+      "step": 1012
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4130177542467055e-05,
+      "loss": 0.3192,
+      "step": 1013
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4118702253484155e-05,
+      "loss": 0.3599,
+      "step": 1014
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4107220429176301e-05,
+      "loss": 0.2313,
+      "step": 1015
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4095732087762199e-05,
+      "loss": 0.2029,
+      "step": 1016
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.4084237247470907e-05,
+      "loss": 0.2828,
+      "step": 1017
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4072735926541788e-05,
+      "loss": 0.269,
+      "step": 1018
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4061228143224493e-05,
+      "loss": 0.2322,
+      "step": 1019
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4049713915778918e-05,
+      "loss": 0.2849,
+      "step": 1020
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.40381932624752e-05,
+      "loss": 0.344,
+      "step": 1021
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4026666201593661e-05,
+      "loss": 0.2516,
+      "step": 1022
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.401513275142479e-05,
+      "loss": 0.2141,
+      "step": 1023
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.4003592930269219e-05,
+      "loss": 0.2684,
+      "step": 1024
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3992046756437685e-05,
+      "loss": 0.3212,
+      "step": 1025
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.398049424825101e-05,
+      "loss": 0.342,
+      "step": 1026
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.396893542404006e-05,
+      "loss": 0.3219,
+      "step": 1027
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3957370302145728e-05,
+      "loss": 0.253,
+      "step": 1028
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.39457989009189e-05,
+      "loss": 0.2881,
+      "step": 1029
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3934221238720422e-05,
+      "loss": 0.3058,
+      "step": 1030
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3922637333921082e-05,
+      "loss": 0.1972,
+      "step": 1031
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.391104720490156e-05,
+      "loss": 0.2199,
+      "step": 1032
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3899450870052431e-05,
+      "loss": 0.2496,
+      "step": 1033
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3887848347774095e-05,
+      "loss": 0.2284,
+      "step": 1034
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3876239656476788e-05,
+      "loss": 0.2701,
+      "step": 1035
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.386462481458053e-05,
+      "loss": 0.355,
+      "step": 1036
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3853003840515094e-05,
+      "loss": 0.2597,
+      "step": 1037
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3841376752719988e-05,
+      "loss": 0.3126,
+      "step": 1038
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3829743569644423e-05,
+      "loss": 0.2496,
+      "step": 1039
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.381810430974728e-05,
+      "loss": 0.2383,
+      "step": 1040
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3806458991497079e-05,
+      "loss": 0.2985,
+      "step": 1041
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3794807633371954e-05,
+      "loss": 0.2693,
+      "step": 1042
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.378315025385963e-05,
+      "loss": 0.3073,
+      "step": 1043
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3771486871457376e-05,
+      "loss": 0.2787,
+      "step": 1044
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3759817504671991e-05,
+      "loss": 0.3053,
+      "step": 1045
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3748142172019772e-05,
+      "loss": 0.2936,
+      "step": 1046
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3736460892026482e-05,
+      "loss": 0.3285,
+      "step": 1047
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3724773683227313e-05,
+      "loss": 0.3097,
+      "step": 1048
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3713080564166873e-05,
+      "loss": 0.2331,
+      "step": 1049
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3701381553399147e-05,
+      "loss": 0.2677,
+      "step": 1050
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3689676669487462e-05,
+      "loss": 0.2507,
+      "step": 1051
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3677965931004471e-05,
+      "loss": 0.3398,
+      "step": 1052
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3666249356532123e-05,
+      "loss": 0.257,
+      "step": 1053
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3654526964661606e-05,
+      "loss": 0.2813,
+      "step": 1054
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3642798773993358e-05,
+      "loss": 0.2349,
+      "step": 1055
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3631064803137014e-05,
+      "loss": 0.2526,
+      "step": 1056
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3619325070711376e-05,
+      "loss": 0.244,
+      "step": 1057
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3607579595344389e-05,
+      "loss": 0.2098,
+      "step": 1058
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3595828395673116e-05,
+      "loss": 0.2676,
+      "step": 1059
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3584071490343697e-05,
+      "loss": 0.2084,
+      "step": 1060
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3572308898011328e-05,
+      "loss": 0.3086,
+      "step": 1061
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3560540637340232e-05,
+      "loss": 0.2661,
+      "step": 1062
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3548766727003617e-05,
+      "loss": 0.2964,
+      "step": 1063
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3536987185683668e-05,
+      "loss": 0.2552,
+      "step": 1064
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3525202032071492e-05,
+      "loss": 0.2434,
+      "step": 1065
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3513411284867108e-05,
+      "loss": 0.2323,
+      "step": 1066
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3501614962779415e-05,
+      "loss": 0.264,
+      "step": 1067
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.348981308452615e-05,
+      "loss": 0.2385,
+      "step": 1068
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3478005668833869e-05,
+      "loss": 0.3053,
+      "step": 1069
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3466192734437914e-05,
+      "loss": 0.227,
+      "step": 1070
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.345437430008239e-05,
+      "loss": 0.306,
+      "step": 1071
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.344255038452012e-05,
+      "loss": 0.2765,
+      "step": 1072
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3430721006512622e-05,
+      "loss": 0.3679,
+      "step": 1073
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3418886184830098e-05,
+      "loss": 0.2584,
+      "step": 1074
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3407045938251369e-05,
+      "loss": 0.2765,
+      "step": 1075
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.339520028556387e-05,
+      "loss": 0.2509,
+      "step": 1076
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3383349245563628e-05,
+      "loss": 0.2309,
+      "step": 1077
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3371492837055196e-05,
+      "loss": 0.1647,
+      "step": 1078
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3359631078851659e-05,
+      "loss": 0.2757,
+      "step": 1079
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3347763989774585e-05,
+      "loss": 0.2258,
+      "step": 1080
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3335891588654008e-05,
+      "loss": 0.2927,
+      "step": 1081
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3324013894328384e-05,
+      "loss": 0.2264,
+      "step": 1082
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3312130925644569e-05,
+      "loss": 0.327,
+      "step": 1083
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3300242701457789e-05,
+      "loss": 0.2717,
+      "step": 1084
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3288349240631608e-05,
+      "loss": 0.2897,
+      "step": 1085
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3276450562037903e-05,
+      "loss": 0.2526,
+      "step": 1086
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3264546684556824e-05,
+      "loss": 0.3415,
+      "step": 1087
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3252637627076784e-05,
+      "loss": 0.2715,
+      "step": 1088
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3240723408494391e-05,
+      "loss": 0.2898,
+      "step": 1089
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3228804047714462e-05,
+      "loss": 0.247,
+      "step": 1090
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3216879563649973e-05,
+      "loss": 0.2417,
+      "step": 1091
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.320494997522202e-05,
+      "loss": 0.255,
+      "step": 1092
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.31930153013598e-05,
+      "loss": 0.3065,
+      "step": 1093
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3181075561000587e-05,
+      "loss": 0.2648,
+      "step": 1094
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.316913077308969e-05,
+      "loss": 0.3068,
+      "step": 1095
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3157180956580418e-05,
+      "loss": 0.2678,
+      "step": 1096
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3145226130434075e-05,
+      "loss": 0.3062,
+      "step": 1097
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.31332663136199e-05,
+      "loss": 0.2811,
+      "step": 1098
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3121301525115069e-05,
+      "loss": 0.3061,
+      "step": 1099
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.310933178390462e-05,
+      "loss": 0.25,
+      "step": 1100
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3097357108981475e-05,
+      "loss": 0.3051,
+      "step": 1101
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3085377519346372e-05,
+      "loss": 0.3557,
+      "step": 1102
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.3073393034007848e-05,
+      "loss": 0.2938,
+      "step": 1103
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.306140367198221e-05,
+      "loss": 0.2783,
+      "step": 1104
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3049409452293507e-05,
+      "loss": 0.2514,
+      "step": 1105
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3037410393973486e-05,
+      "loss": 0.2629,
+      "step": 1106
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3025406516061581e-05,
+      "loss": 0.2719,
+      "step": 1107
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3013397837604869e-05,
+      "loss": 0.261,
+      "step": 1108
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.3001384377658046e-05,
+      "loss": 0.3257,
+      "step": 1109
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2989366155283394e-05,
+      "loss": 0.3044,
+      "step": 1110
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.297734318955075e-05,
+      "loss": 0.2107,
+      "step": 1111
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2965315499537483e-05,
+      "loss": 0.2671,
+      "step": 1112
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2953283104328454e-05,
+      "loss": 0.2668,
+      "step": 1113
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2941246023015986e-05,
+      "loss": 0.1965,
+      "step": 1114
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2929204274699845e-05,
+      "loss": 0.2609,
+      "step": 1115
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.29171578784872e-05,
+      "loss": 0.2547,
+      "step": 1116
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2905106853492598e-05,
+      "loss": 0.2427,
+      "step": 1117
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2893051218837922e-05,
+      "loss": 0.3239,
+      "step": 1118
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2880990993652379e-05,
+      "loss": 0.2668,
+      "step": 1119
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2868926197072452e-05,
+      "loss": 0.2667,
+      "step": 1120
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2856856848241879e-05,
+      "loss": 0.3071,
+      "step": 1121
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2844782966311629e-05,
+      "loss": 0.2575,
+      "step": 1122
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.283270457043986e-05,
+      "loss": 0.2329,
+      "step": 1123
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2820621679791882e-05,
+      "loss": 0.2936,
+      "step": 1124
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2808534313540153e-05,
+      "loss": 0.2975,
+      "step": 1125
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2796442490864227e-05,
+      "loss": 0.2785,
+      "step": 1126
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2784346230950724e-05,
+      "loss": 0.2456,
+      "step": 1127
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2772245552993306e-05,
+      "loss": 0.2703,
+      "step": 1128
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2760140476192656e-05,
+      "loss": 0.3245,
+      "step": 1129
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2748031019756417e-05,
+      "loss": 0.2592,
+      "step": 1130
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2735917202899203e-05,
+      "loss": 0.343,
+      "step": 1131
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2723799044842533e-05,
+      "loss": 0.2898,
+      "step": 1132
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2711676564814818e-05,
+      "loss": 0.2441,
+      "step": 1133
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.269954978205133e-05,
+      "loss": 0.2991,
+      "step": 1134
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2687418715794161e-05,
+      "loss": 0.2485,
+      "step": 1135
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2675283385292212e-05,
+      "loss": 0.3002,
+      "step": 1136
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2663143809801133e-05,
+      "loss": 0.3144,
+      "step": 1137
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.265100000858332e-05,
+      "loss": 0.3256,
+      "step": 1138
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2638852000907883e-05,
+      "loss": 0.2744,
+      "step": 1139
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2626699806050586e-05,
+      "loss": 0.2762,
+      "step": 1140
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2614543443293856e-05,
+      "loss": 0.2767,
+      "step": 1141
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.260238293192672e-05,
+      "loss": 0.2092,
+      "step": 1142
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2590218291244797e-05,
+      "loss": 0.2215,
+      "step": 1143
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2578049540550251e-05,
+      "loss": 0.2179,
+      "step": 1144
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2565876699151773e-05,
+      "loss": 0.2561,
+      "step": 1145
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2553699786364541e-05,
+      "loss": 0.2598,
+      "step": 1146
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2541518821510199e-05,
+      "loss": 0.1917,
+      "step": 1147
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2529333823916807e-05,
+      "loss": 0.2666,
+      "step": 1148
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2517144812918844e-05,
+      "loss": 0.2806,
+      "step": 1149
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2504951807857144e-05,
+      "loss": 0.2962,
+      "step": 1150
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2492754828078878e-05,
+      "loss": 0.2649,
+      "step": 1151
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.248055389293753e-05,
+      "loss": 0.331,
+      "step": 1152
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2468349021792856e-05,
+      "loss": 0.282,
+      "step": 1153
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.245614023401086e-05,
+      "loss": 0.3312,
+      "step": 1154
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2443927548963757e-05,
+      "loss": 0.2865,
+      "step": 1155
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2431710986029948e-05,
+      "loss": 0.2487,
+      "step": 1156
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2419490564593995e-05,
+      "loss": 0.2875,
+      "step": 1157
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2407266304046565e-05,
+      "loss": 0.2529,
+      "step": 1158
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2395038223784432e-05,
+      "loss": 0.2338,
+      "step": 1159
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2382806343210428e-05,
+      "loss": 0.2846,
+      "step": 1160
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2370570681733405e-05,
+      "loss": 0.2804,
+      "step": 1161
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2358331258768228e-05,
+      "loss": 0.3026,
+      "step": 1162
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2346088093735725e-05,
+      "loss": 0.2244,
+      "step": 1163
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2333841206062656e-05,
+      "loss": 0.2733,
+      "step": 1164
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.23215906151817e-05,
+      "loss": 0.2594,
+      "step": 1165
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.23093363405314e-05,
+      "loss": 0.2966,
+      "step": 1166
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2297078401556153e-05,
+      "loss": 0.2009,
+      "step": 1167
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2284816817706165e-05,
+      "loss": 0.2796,
+      "step": 1168
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2272551608437426e-05,
+      "loss": 0.2393,
+      "step": 1169
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2260282793211682e-05,
+      "loss": 0.2323,
+      "step": 1170
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2248010391496398e-05,
+      "loss": 0.2141,
+      "step": 1171
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2235734422764729e-05,
+      "loss": 0.2913,
+      "step": 1172
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2223454906495494e-05,
+      "loss": 0.291,
+      "step": 1173
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2211171862173136e-05,
+      "loss": 0.2781,
+      "step": 1174
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2198885309287701e-05,
+      "loss": 0.2381,
+      "step": 1175
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2186595267334795e-05,
+      "loss": 0.2949,
+      "step": 1176
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2174301755815572e-05,
+      "loss": 0.2703,
+      "step": 1177
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.216200479423668e-05,
+      "loss": 0.2811,
+      "step": 1178
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2149704402110243e-05,
+      "loss": 0.2714,
+      "step": 1179
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2137400598953833e-05,
+      "loss": 0.3351,
+      "step": 1180
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2125093404290438e-05,
+      "loss": 0.3007,
+      "step": 1181
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2112782837648416e-05,
+      "loss": 0.2702,
+      "step": 1182
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2100468918561481e-05,
+      "loss": 0.2117,
+      "step": 1183
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2088151666568671e-05,
+      "loss": 0.2921,
+      "step": 1184
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2075831101214307e-05,
+      "loss": 0.2411,
+      "step": 1185
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2063507242047962e-05,
+      "loss": 0.3394,
+      "step": 1186
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2051180108624452e-05,
+      "loss": 0.2952,
+      "step": 1187
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2038849720503773e-05,
+      "loss": 0.2952,
+      "step": 1188
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2026516097251095e-05,
+      "loss": 0.2993,
+      "step": 1189
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2014179258436713e-05,
+      "loss": 0.2374,
+      "step": 1190
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2001839223636032e-05,
+      "loss": 0.2783,
+      "step": 1191
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1989496012429522e-05,
+      "loss": 0.3058,
+      "step": 1192
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1977149644402697e-05,
+      "loss": 0.3334,
+      "step": 1193
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.196480013914608e-05,
+      "loss": 0.284,
+      "step": 1194
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.195244751625517e-05,
+      "loss": 0.2615,
+      "step": 1195
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.194009179533041e-05,
+      "loss": 0.2928,
+      "step": 1196
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.192773299597717e-05,
+      "loss": 0.2537,
+      "step": 1197
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1915371137805698e-05,
+      "loss": 0.2902,
+      "step": 1198
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1903006240431086e-05,
+      "loss": 0.2907,
+      "step": 1199
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1890638323473263e-05,
+      "loss": 0.3383,
+      "step": 1200
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1878267406556942e-05,
+      "loss": 0.2829,
+      "step": 1201
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1865893509311604e-05,
+      "loss": 0.2615,
+      "step": 1202
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1853516651371441e-05,
+      "loss": 0.2418,
+      "step": 1203
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1841136852375361e-05,
+      "loss": 0.237,
+      "step": 1204
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1828754131966937e-05,
+      "loss": 0.2225,
+      "step": 1205
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1816368509794365e-05,
+      "loss": 0.2677,
+      "step": 1206
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1803980005510459e-05,
+      "loss": 0.266,
+      "step": 1207
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1791588638772595e-05,
+      "loss": 0.3242,
+      "step": 1208
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1779194429242704e-05,
+      "loss": 0.2946,
+      "step": 1209
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1766797396587213e-05,
+      "loss": 0.2227,
+      "step": 1210
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.175439756047704e-05,
+      "loss": 0.2141,
+      "step": 1211
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1741994940587544e-05,
+      "loss": 0.2547,
+      "step": 1212
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1729589556598506e-05,
+      "loss": 0.2884,
+      "step": 1213
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1717181428194091e-05,
+      "loss": 0.2711,
+      "step": 1214
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1704770575062823e-05,
+      "loss": 0.2934,
+      "step": 1215
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1692357016897537e-05,
+      "loss": 0.2505,
+      "step": 1216
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1679940773395375e-05,
+      "loss": 0.263,
+      "step": 1217
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1667521864257729e-05,
+      "loss": 0.4028,
+      "step": 1218
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1655100309190232e-05,
+      "loss": 0.2859,
+      "step": 1219
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1642676127902699e-05,
+      "loss": 0.2658,
+      "step": 1220
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1630249340109127e-05,
+      "loss": 0.2832,
+      "step": 1221
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.161781996552765e-05,
+      "loss": 0.3072,
+      "step": 1222
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1605388023880488e-05,
+      "loss": 0.2752,
+      "step": 1223
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1592953534893954e-05,
+      "loss": 0.2769,
+      "step": 1224
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.158051651829839e-05,
+      "loss": 0.2166,
+      "step": 1225
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1568076993828163e-05,
+      "loss": 0.1676,
+      "step": 1226
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1555634981221602e-05,
+      "loss": 0.2637,
+      "step": 1227
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1543190500220991e-05,
+      "loss": 0.2598,
+      "step": 1228
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1530743570572542e-05,
+      "loss": 0.2004,
+      "step": 1229
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1518294212026336e-05,
+      "loss": 0.2291,
+      "step": 1230
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1505842444336316e-05,
+      "loss": 0.2975,
+      "step": 1231
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1493388287260241e-05,
+      "loss": 0.2926,
+      "step": 1232
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1480931760559678e-05,
+      "loss": 0.3296,
+      "step": 1233
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.146847288399993e-05,
+      "loss": 0.2816,
+      "step": 1234
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1456011677350052e-05,
+      "loss": 0.2815,
+      "step": 1235
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1443548160382776e-05,
+      "loss": 0.304,
+      "step": 1236
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1431082352874516e-05,
+      "loss": 0.2737,
+      "step": 1237
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1418614274605311e-05,
+      "loss": 0.3106,
+      "step": 1238
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1406143945358807e-05,
+      "loss": 0.2841,
+      "step": 1239
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1393671384922222e-05,
+      "loss": 0.241,
+      "step": 1240
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.138119661308631e-05,
+      "loss": 0.2659,
+      "step": 1241
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.136871964964534e-05,
+      "loss": 0.2458,
+      "step": 1242
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1356240514397055e-05,
+      "loss": 0.2417,
+      "step": 1243
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1343759227142636e-05,
+      "loss": 0.2906,
+      "step": 1244
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1331275807686701e-05,
+      "loss": 0.1998,
+      "step": 1245
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.131879027583723e-05,
+      "loss": 0.2597,
+      "step": 1246
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1306302651405563e-05,
+      "loss": 0.3295,
+      "step": 1247
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.129381295420636e-05,
+      "loss": 0.2842,
+      "step": 1248
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.128132120405757e-05,
+      "loss": 0.2316,
+      "step": 1249
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.12688274207804e-05,
+      "loss": 0.2626,
+      "step": 1250
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.125633162419928e-05,
+      "loss": 0.2108,
+      "step": 1251
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1243833834141835e-05,
+      "loss": 0.3241,
+      "step": 1252
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1231334070438864e-05,
+      "loss": 0.2564,
+      "step": 1253
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1218832352924278e-05,
+      "loss": 0.2991,
+      "step": 1254
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1206328701435105e-05,
+      "loss": 0.3217,
+      "step": 1255
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1193823135811434e-05,
+      "loss": 0.2422,
+      "step": 1256
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1181315675896393e-05,
+      "loss": 0.2078,
+      "step": 1257
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1168806341536116e-05,
+      "loss": 0.1969,
+      "step": 1258
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.115629515257971e-05,
+      "loss": 0.293,
+      "step": 1259
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1143782128879223e-05,
+      "loss": 0.231,
+      "step": 1260
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1131267290289622e-05,
+      "loss": 0.2613,
+      "step": 1261
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1118750656668746e-05,
+      "loss": 0.216,
+      "step": 1262
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1106232247877285e-05,
+      "loss": 0.2546,
+      "step": 1263
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1093712083778748e-05,
+      "loss": 0.1622,
+      "step": 1264
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1081190184239418e-05,
+      "loss": 0.2245,
+      "step": 1265
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1068666569128352e-05,
+      "loss": 0.2221,
+      "step": 1266
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1056141258317312e-05,
+      "loss": 0.197,
+      "step": 1267
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.1043614271680756e-05,
+      "loss": 0.2725,
+      "step": 1268
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.10310856290958e-05,
+      "loss": 0.3346,
+      "step": 1269
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1018555350442198e-05,
+      "loss": 0.2567,
+      "step": 1270
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.1006023455602279e-05,
+      "loss": 0.218,
+      "step": 1271
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0993489964460957e-05,
+      "loss": 0.2088,
+      "step": 1272
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0980954896905667e-05,
+      "loss": 0.2047,
+      "step": 1273
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0968418272826348e-05,
+      "loss": 0.2658,
+      "step": 1274
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0955880112115412e-05,
+      "loss": 0.2828,
+      "step": 1275
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0943340434667701e-05,
+      "loss": 0.2586,
+      "step": 1276
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0930799260380482e-05,
+      "loss": 0.2515,
+      "step": 1277
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0918256609153373e-05,
+      "loss": 0.2289,
+      "step": 1278
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.090571250088835e-05,
+      "loss": 0.2675,
+      "step": 1279
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.08931669554897e-05,
+      "loss": 0.3212,
+      "step": 1280
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0880619992863988e-05,
+      "loss": 0.2925,
+      "step": 1281
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0868071632920027e-05,
+      "loss": 0.2426,
+      "step": 1282
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0855521895568849e-05,
+      "loss": 0.2581,
+      "step": 1283
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0842970800723671e-05,
+      "loss": 0.2851,
+      "step": 1284
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0830418368299864e-05,
+      "loss": 0.2877,
+      "step": 1285
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0817864618214921e-05,
+      "loss": 0.2663,
+      "step": 1286
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0805309570388427e-05,
+      "loss": 0.244,
+      "step": 1287
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0792753244742028e-05,
+      "loss": 0.2232,
+      "step": 1288
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0780195661199389e-05,
+      "loss": 0.2953,
+      "step": 1289
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.076763683968618e-05,
+      "loss": 0.2592,
+      "step": 1290
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0755076800130032e-05,
+      "loss": 0.2844,
+      "step": 1291
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.074251556246051e-05,
+      "loss": 0.3364,
+      "step": 1292
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0729953146609076e-05,
+      "loss": 0.2694,
+      "step": 1293
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0717389572509068e-05,
+      "loss": 0.2264,
+      "step": 1294
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.070482486009566e-05,
+      "loss": 0.2403,
+      "step": 1295
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0692259029305823e-05,
+      "loss": 0.2948,
+      "step": 1296
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0679692100078317e-05,
+      "loss": 0.2093,
+      "step": 1297
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0667124092353636e-05,
+      "loss": 0.2666,
+      "step": 1298
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0654555026073986e-05,
+      "loss": 0.2417,
+      "step": 1299
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0641984921183255e-05,
+      "loss": 0.27,
+      "step": 1300
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.062941379762698e-05,
+      "loss": 0.2845,
+      "step": 1301
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.061684167535231e-05,
+      "loss": 0.3337,
+      "step": 1302
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0604268574307981e-05,
+      "loss": 0.3486,
+      "step": 1303
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0591694514444284e-05,
+      "loss": 0.2472,
+      "step": 1304
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.057911951571303e-05,
+      "loss": 0.2631,
+      "step": 1305
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0566543598067518e-05,
+      "loss": 0.2087,
+      "step": 1306
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0553966781462503e-05,
+      "loss": 0.2314,
+      "step": 1307
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0541389085854177e-05,
+      "loss": 0.2358,
+      "step": 1308
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0528810531200113e-05,
+      "loss": 0.2854,
+      "step": 1309
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0516231137459259e-05,
+      "loss": 0.2383,
+      "step": 1310
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0503650924591886e-05,
+      "loss": 0.2822,
+      "step": 1311
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.049106991255957e-05,
+      "loss": 0.2311,
+      "step": 1312
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.047848812132515e-05,
+      "loss": 0.2825,
+      "step": 1313
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0465905570852706e-05,
+      "loss": 0.2648,
+      "step": 1314
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0453322281107518e-05,
+      "loss": 0.2513,
+      "step": 1315
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0440738272056048e-05,
+      "loss": 0.2627,
+      "step": 1316
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0428153563665889e-05,
+      "loss": 0.2301,
+      "step": 1317
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0415568175905748e-05,
+      "loss": 0.2204,
+      "step": 1318
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0402982128745416e-05,
+      "loss": 0.3054,
+      "step": 1319
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0390395442155716e-05,
+      "loss": 0.2761,
+      "step": 1320
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0377808136108495e-05,
+      "loss": 0.2498,
+      "step": 1321
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0365220230576592e-05,
+      "loss": 0.2961,
+      "step": 1322
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0352631745533775e-05,
+      "loss": 0.2795,
+      "step": 1323
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0340042700954747e-05,
+      "loss": 0.2633,
+      "step": 1324
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0327453116815099e-05,
+      "loss": 0.2329,
+      "step": 1325
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0314863013091273e-05,
+      "loss": 0.2221,
+      "step": 1326
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0302272409760535e-05,
+      "loss": 0.2429,
+      "step": 1327
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0289681326800946e-05,
+      "loss": 0.2567,
+      "step": 1328
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0277089784191334e-05,
+      "loss": 0.3415,
+      "step": 1329
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0264497801911241e-05,
+      "loss": 0.284,
+      "step": 1330
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0251905399940922e-05,
+      "loss": 0.2419,
+      "step": 1331
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.023931259826129e-05,
+      "loss": 0.3027,
+      "step": 1332
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0226719416853894e-05,
+      "loss": 0.2195,
+      "step": 1333
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0214125875700887e-05,
+      "loss": 0.2143,
+      "step": 1334
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0201531994784987e-05,
+      "loss": 0.2786,
+      "step": 1335
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0188937794089464e-05,
+      "loss": 0.2445,
+      "step": 1336
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.017634329359808e-05,
+      "loss": 0.2521,
+      "step": 1337
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.016374851329508e-05,
+      "loss": 0.2064,
+      "step": 1338
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0151153473165154e-05,
+      "loss": 0.2165,
+      "step": 1339
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0138558193193402e-05,
+      "loss": 0.3081,
+      "step": 1340
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0125962693365308e-05,
+      "loss": 0.2558,
+      "step": 1341
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.01133669936667e-05,
+      "loss": 0.305,
+      "step": 1342
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0100771114083727e-05,
+      "loss": 0.236,
+      "step": 1343
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0088175074602823e-05,
+      "loss": 0.2112,
+      "step": 1344
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0075578895210667e-05,
+      "loss": 0.2727,
+      "step": 1345
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.0062982595894176e-05,
+      "loss": 0.2947,
+      "step": 1346
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0050386196640447e-05,
+      "loss": 0.2941,
+      "step": 1347
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0037789717436734e-05,
+      "loss": 0.2353,
+      "step": 1348
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0025193178270428e-05,
+      "loss": 0.2507,
+      "step": 1349
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.0012596599129003e-05,
+      "loss": 0.2324,
+      "step": 1350
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1e-05,
+      "loss": 0.2934,
+      "step": 1351
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987403400871e-06,
+      "loss": 0.2517,
+      "step": 1352
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.974806821729577e-06,
+      "loss": 0.2443,
+      "step": 1353
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.962210282563268e-06,
+      "loss": 0.3035,
+      "step": 1354
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.949613803359557e-06,
+      "loss": 0.1949,
+      "step": 1355
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.937017404105826e-06,
+      "loss": 0.1929,
+      "step": 1356
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.924421104789338e-06,
+      "loss": 0.254,
+      "step": 1357
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.911824925397184e-06,
+      "loss": 0.3034,
+      "step": 1358
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.899228885916276e-06,
+      "loss": 0.2518,
+      "step": 1359
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.8866330063333e-06,
+      "loss": 0.2174,
+      "step": 1360
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.874037306634692e-06,
+      "loss": 0.2228,
+      "step": 1361
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.861441806806598e-06,
+      "loss": 0.287,
+      "step": 1362
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.84884652683485e-06,
+      "loss": 0.205,
+      "step": 1363
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.836251486704923e-06,
+      "loss": 0.274,
+      "step": 1364
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.823656706401922e-06,
+      "loss": 0.304,
+      "step": 1365
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.811062205910538e-06,
+      "loss": 0.2407,
+      "step": 1366
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.798468005215014e-06,
+      "loss": 0.2882,
+      "step": 1367
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.785874124299116e-06,
+      "loss": 0.2315,
+      "step": 1368
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.77328058314611e-06,
+      "loss": 0.226,
+      "step": 1369
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.760687401738712e-06,
+      "loss": 0.3053,
+      "step": 1370
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.748094600059083e-06,
+      "loss": 0.2633,
+      "step": 1371
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.735502198088764e-06,
+      "loss": 0.1746,
+      "step": 1372
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.722910215808673e-06,
+      "loss": 0.3572,
+      "step": 1373
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.710318673199056e-06,
+      "loss": 0.2783,
+      "step": 1374
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.697727590239467e-06,
+      "loss": 0.2012,
+      "step": 1375
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.685136986908729e-06,
+      "loss": 0.2064,
+      "step": 1376
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.672546883184903e-06,
+      "loss": 0.2991,
+      "step": 1377
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.659957299045254e-06,
+      "loss": 0.3332,
+      "step": 1378
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.647368254466229e-06,
+      "loss": 0.2444,
+      "step": 1379
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.634779769423412e-06,
+      "loss": 0.3233,
+      "step": 1380
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.622191863891506e-06,
+      "loss": 0.2864,
+      "step": 1381
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.609604557844287e-06,
+      "loss": 0.2732,
+      "step": 1382
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.597017871254588e-06,
+      "loss": 0.2464,
+      "step": 1383
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.584431824094254e-06,
+      "loss": 0.2487,
+      "step": 1384
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.571846436334115e-06,
+      "loss": 0.2762,
+      "step": 1385
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.559261727943955e-06,
+      "loss": 0.307,
+      "step": 1386
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.546677718892485e-06,
+      "loss": 0.2302,
+      "step": 1387
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.5340944291473e-06,
+      "loss": 0.2629,
+      "step": 1388
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.521511878674854e-06,
+      "loss": 0.2121,
+      "step": 1389
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.508930087440436e-06,
+      "loss": 0.1837,
+      "step": 1390
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.496349075408114e-06,
+      "loss": 0.2175,
+      "step": 1391
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.483768862540741e-06,
+      "loss": 0.2839,
+      "step": 1392
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.471189468799887e-06,
+      "loss": 0.3442,
+      "step": 1393
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.458610914145826e-06,
+      "loss": 0.3584,
+      "step": 1394
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.446033218537498e-06,
+      "loss": 0.26,
+      "step": 1395
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.433456401932485e-06,
+      "loss": 0.2798,
+      "step": 1396
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.420880484286971e-06,
+      "loss": 0.2105,
+      "step": 1397
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.40830548555572e-06,
+      "loss": 0.2015,
+      "step": 1398
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.395731425692022e-06,
+      "loss": 0.3099,
+      "step": 1399
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.383158324647693e-06,
+      "loss": 0.2488,
+      "step": 1400
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.370586202373023e-06,
+      "loss": 0.2436,
+      "step": 1401
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.35801507881675e-06,
+      "loss": 0.3674,
+      "step": 1402
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.345444973926019e-06,
+      "loss": 0.3363,
+      "step": 1403
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.332875907646369e-06,
+      "loss": 0.215,
+      "step": 1404
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.320307899921688e-06,
+      "loss": 0.2057,
+      "step": 1405
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.307740970694182e-06,
+      "loss": 0.2137,
+      "step": 1406
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.295175139904342e-06,
+      "loss": 0.2519,
+      "step": 1407
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.282610427490934e-06,
+      "loss": 0.2828,
+      "step": 1408
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.270046853390924e-06,
+      "loss": 0.3547,
+      "step": 1409
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.257484437539492e-06,
+      "loss": 0.2104,
+      "step": 1410
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.24492319986997e-06,
+      "loss": 0.2023,
+      "step": 1411
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.232363160313823e-06,
+      "loss": 0.2916,
+      "step": 1412
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.219804338800614e-06,
+      "loss": 0.3083,
+      "step": 1413
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.207246755257974e-06,
+      "loss": 0.1902,
+      "step": 1414
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.194690429611577e-06,
+      "loss": 0.2717,
+      "step": 1415
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.182135381785082e-06,
+      "loss": 0.325,
+      "step": 1416
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.169581631700139e-06,
+      "loss": 0.216,
+      "step": 1417
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.157029199276332e-06,
+      "loss": 0.2036,
+      "step": 1418
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.144478104431156e-06,
+      "loss": 0.2008,
+      "step": 1419
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.131928367079978e-06,
+      "loss": 0.29,
+      "step": 1420
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.119380007136017e-06,
+      "loss": 0.2169,
+      "step": 1421
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.106833044510303e-06,
+      "loss": 0.3744,
+      "step": 1422
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.094287499111652e-06,
+      "loss": 0.2411,
+      "step": 1423
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.081743390846629e-06,
+      "loss": 0.3203,
+      "step": 1424
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.069200739619521e-06,
+      "loss": 0.1375,
+      "step": 1425
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.056659565332298e-06,
+      "loss": 0.3054,
+      "step": 1426
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.04411988788459e-06,
+      "loss": 0.2619,
+      "step": 1427
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.031581727173653e-06,
+      "loss": 0.2336,
+      "step": 1428
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.019045103094338e-06,
+      "loss": 0.3196,
+      "step": 1429
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.006510035539046e-06,
+      "loss": 0.3034,
+      "step": 1430
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.993976544397723e-06,
+      "loss": 0.342,
+      "step": 1431
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.981444649557806e-06,
+      "loss": 0.2655,
+      "step": 1432
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.968914370904201e-06,
+      "loss": 0.2148,
+      "step": 1433
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.956385728319249e-06,
+      "loss": 0.2477,
+      "step": 1434
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.943858741682692e-06,
+      "loss": 0.3527,
+      "step": 1435
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.931333430871654e-06,
+      "loss": 0.2489,
+      "step": 1436
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.918809815760585e-06,
+      "loss": 0.326,
+      "step": 1437
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.906287916221259e-06,
+      "loss": 0.2695,
+      "step": 1438
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.893767752122717e-06,
+      "loss": 0.2232,
+      "step": 1439
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.881249343331256e-06,
+      "loss": 0.2373,
+      "step": 1440
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.868732709710378e-06,
+      "loss": 0.2381,
+      "step": 1441
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.856217871120779e-06,
+      "loss": 0.1834,
+      "step": 1442
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.843704847420294e-06,
+      "loss": 0.3,
+      "step": 1443
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.831193658463886e-06,
+      "loss": 0.2817,
+      "step": 1444
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.818684324103608e-06,
+      "loss": 0.2657,
+      "step": 1445
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.80617686418857e-06,
+      "loss": 0.2265,
+      "step": 1446
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.793671298564898e-06,
+      "loss": 0.2813,
+      "step": 1447
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.781167647075724e-06,
+      "loss": 0.2381,
+      "step": 1448
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.76866592956114e-06,
+      "loss": 0.2573,
+      "step": 1449
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.756166165858168e-06,
+      "loss": 0.254,
+      "step": 1450
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.743668375800724e-06,
+      "loss": 0.2029,
+      "step": 1451
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.731172579219605e-06,
+      "loss": 0.351,
+      "step": 1452
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.718678795942434e-06,
+      "loss": 0.2503,
+      "step": 1453
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.706187045793646e-06,
+      "loss": 0.2732,
+      "step": 1454
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.693697348594437e-06,
+      "loss": 0.2641,
+      "step": 1455
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.681209724162772e-06,
+      "loss": 0.2542,
+      "step": 1456
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.6687241923133e-06,
+      "loss": 0.2452,
+      "step": 1457
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.656240772857364e-06,
+      "loss": 0.2494,
+      "step": 1458
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.64375948560295e-06,
+      "loss": 0.2953,
+      "step": 1459
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.631280350354664e-06,
+      "loss": 0.2171,
+      "step": 1460
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.618803386913692e-06,
+      "loss": 0.2027,
+      "step": 1461
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.60632861507778e-06,
+      "loss": 0.2318,
+      "step": 1462
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.593856054641195e-06,
+      "loss": 0.2677,
+      "step": 1463
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.58138572539469e-06,
+      "loss": 0.2405,
+      "step": 1464
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.568917647125487e-06,
+      "loss": 0.3167,
+      "step": 1465
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.556451839617227e-06,
+      "loss": 0.2887,
+      "step": 1466
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.543988322649954e-06,
+      "loss": 0.2332,
+      "step": 1467
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.531527116000073e-06,
+      "loss": 0.2648,
+      "step": 1468
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.519068239440328e-06,
+      "loss": 0.2473,
+      "step": 1469
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.50661171273976e-06,
+      "loss": 0.3028,
+      "step": 1470
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.494157555663686e-06,
+      "loss": 0.3213,
+      "step": 1471
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.481705787973666e-06,
+      "loss": 0.2852,
+      "step": 1472
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.46925642942746e-06,
+      "loss": 0.2506,
+      "step": 1473
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.456809499779008e-06,
+      "loss": 0.1896,
+      "step": 1474
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.444365018778401e-06,
+      "loss": 0.2056,
+      "step": 1475
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.431923006171839e-06,
+      "loss": 0.3152,
+      "step": 1476
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.419483481701611e-06,
+      "loss": 0.2278,
+      "step": 1477
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.40704646510605e-06,
+      "loss": 0.2006,
+      "step": 1478
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.394611976119514e-06,
+      "loss": 0.2933,
+      "step": 1479
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.382180034472353e-06,
+      "loss": 0.2837,
+      "step": 1480
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.369750659890874e-06,
+      "loss": 0.2693,
+      "step": 1481
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.357323872097304e-06,
+      "loss": 0.2813,
+      "step": 1482
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.344899690809773e-06,
+      "loss": 0.3538,
+      "step": 1483
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.332478135742273e-06,
+      "loss": 0.2746,
+      "step": 1484
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.32005922660463e-06,
+      "loss": 0.2482,
+      "step": 1485
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.307642983102468e-06,
+      "loss": 0.2645,
+      "step": 1486
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.29522942493718e-06,
+      "loss": 0.2544,
+      "step": 1487
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.282818571805909e-06,
+      "loss": 0.2373,
+      "step": 1488
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.270410443401494e-06,
+      "loss": 0.2607,
+      "step": 1489
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.258005059412458e-06,
+      "loss": 0.2116,
+      "step": 1490
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.245602439522964e-06,
+      "loss": 0.2242,
+      "step": 1491
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.233202603412788e-06,
+      "loss": 0.2599,
+      "step": 1492
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.220805570757298e-06,
+      "loss": 0.1917,
+      "step": 1493
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.208411361227406e-06,
+      "loss": 0.2765,
+      "step": 1494
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.196019994489545e-06,
+      "loss": 0.1976,
+      "step": 1495
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.183631490205636e-06,
+      "loss": 0.1931,
+      "step": 1496
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.171245868033066e-06,
+      "loss": 0.3266,
+      "step": 1497
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.158863147624642e-06,
+      "loss": 0.2938,
+      "step": 1498
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.146483348628564e-06,
+      "loss": 0.2881,
+      "step": 1499
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.134106490688403e-06,
+      "loss": 0.2554,
+      "step": 1500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.12173259344306e-06,
+      "loss": 0.2032,
+      "step": 1501
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.109361676526742e-06,
+      "loss": 0.2473,
+      "step": 1502
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.096993759568914e-06,
+      "loss": 0.1974,
+      "step": 1503
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.084628862194305e-06,
+      "loss": 0.2408,
+      "step": 1504
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.07226700402283e-06,
+      "loss": 0.2227,
+      "step": 1505
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.059908204669588e-06,
+      "loss": 0.2323,
+      "step": 1506
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.047552483744832e-06,
+      "loss": 0.2292,
+      "step": 1507
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.035199860853924e-06,
+      "loss": 0.2614,
+      "step": 1508
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.022850355597305e-06,
+      "loss": 0.2694,
+      "step": 1509
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.010503987570481e-06,
+      "loss": 0.1915,
+      "step": 1510
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.99816077636397e-06,
+      "loss": 0.2087,
+      "step": 1511
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.98582074156329e-06,
+      "loss": 0.2854,
+      "step": 1512
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.973483902748909e-06,
+      "loss": 0.2476,
+      "step": 1513
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.961150279496228e-06,
+      "loss": 0.2329,
+      "step": 1514
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.94881989137555e-06,
+      "loss": 0.2332,
+      "step": 1515
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.936492757952041e-06,
+      "loss": 0.2116,
+      "step": 1516
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.9241688987857e-06,
+      "loss": 0.251,
+      "step": 1517
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.911848333431334e-06,
+      "loss": 0.206,
+      "step": 1518
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.899531081438519e-06,
+      "loss": 0.2675,
+      "step": 1519
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.887217162351585e-06,
+      "loss": 0.2321,
+      "step": 1520
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.874906595709566e-06,
+      "loss": 0.287,
+      "step": 1521
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.862599401046167e-06,
+      "loss": 0.2119,
+      "step": 1522
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.85029559788976e-06,
+      "loss": 0.1794,
+      "step": 1523
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.837995205763324e-06,
+      "loss": 0.2614,
+      "step": 1524
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.825698244184432e-06,
+      "loss": 0.1811,
+      "step": 1525
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.813404732665207e-06,
+      "loss": 0.2261,
+      "step": 1526
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.801114690712302e-06,
+      "loss": 0.226,
+      "step": 1527
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.788828137826866e-06,
+      "loss": 0.204,
+      "step": 1528
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.77654509350451e-06,
+      "loss": 0.2801,
+      "step": 1529
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.764265577235275e-06,
+      "loss": 0.2895,
+      "step": 1530
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.751989608503607e-06,
+      "loss": 0.3547,
+      "step": 1531
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.739717206788321e-06,
+      "loss": 0.3842,
+      "step": 1532
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.727448391562579e-06,
+      "loss": 0.2937,
+      "step": 1533
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.715183182293839e-06,
+      "loss": 0.2321,
+      "step": 1534
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.702921598443849e-06,
+      "loss": 0.2078,
+      "step": 1535
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6906636594686e-06,
+      "loss": 0.2534,
+      "step": 1536
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.678409384818302e-06,
+      "loss": 0.2283,
+      "step": 1537
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.666158793937344e-06,
+      "loss": 0.1906,
+      "step": 1538
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.653911906264279e-06,
+      "loss": 0.2675,
+      "step": 1539
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.641668741231775e-06,
+      "loss": 0.2296,
+      "step": 1540
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.629429318266597e-06,
+      "loss": 0.2669,
+      "step": 1541
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.617193656789576e-06,
+      "loss": 0.2499,
+      "step": 1542
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.604961776215571e-06,
+      "loss": 0.2282,
+      "step": 1543
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.592733695953438e-06,
+      "loss": 0.3051,
+      "step": 1544
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.580509435406009e-06,
+      "loss": 0.2078,
+      "step": 1545
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.568289013970054e-06,
+      "loss": 0.1718,
+      "step": 1546
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.556072451036247e-06,
+      "loss": 0.215,
+      "step": 1547
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.543859765989144e-06,
+      "loss": 0.2672,
+      "step": 1548
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.531650978207147e-06,
+      "loss": 0.2789,
+      "step": 1549
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.519446107062475e-06,
+      "loss": 0.3116,
+      "step": 1550
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.507245171921122e-06,
+      "loss": 0.2297,
+      "step": 1551
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.495048192142858e-06,
+      "loss": 0.2044,
+      "step": 1552
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.482855187081156e-06,
+      "loss": 0.2297,
+      "step": 1553
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.470666176083193e-06,
+      "loss": 0.2416,
+      "step": 1554
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.458481178489805e-06,
+      "loss": 0.2764,
+      "step": 1555
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.446300213635462e-06,
+      "loss": 0.3464,
+      "step": 1556
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.434123300848231e-06,
+      "loss": 0.2208,
+      "step": 1557
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.421950459449752e-06,
+      "loss": 0.2481,
+      "step": 1558
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.409781708755206e-06,
+      "loss": 0.2366,
+      "step": 1559
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3976170680732825e-06,
+      "loss": 0.2763,
+      "step": 1560
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.385456556706148e-06,
+      "loss": 0.2265,
+      "step": 1561
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.373300193949415e-06,
+      "loss": 0.2534,
+      "step": 1562
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.361147999092121e-06,
+      "loss": 0.2296,
+      "step": 1563
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.348999991416683e-06,
+      "loss": 0.2112,
+      "step": 1564
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.336856190198874e-06,
+      "loss": 0.2139,
+      "step": 1565
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.324716614707794e-06,
+      "loss": 0.2725,
+      "step": 1566
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.312581284205839e-06,
+      "loss": 0.2254,
+      "step": 1567
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.300450217948672e-06,
+      "loss": 0.2561,
+      "step": 1568
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.288323435185181e-06,
+      "loss": 0.2967,
+      "step": 1569
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.276200955157469e-06,
+      "loss": 0.2321,
+      "step": 1570
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2640827971008e-06,
+      "loss": 0.2703,
+      "step": 1571
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.2519689802435846e-06,
+      "loss": 0.2799,
+      "step": 1572
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.239859523807349e-06,
+      "loss": 0.2738,
+      "step": 1573
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.227754447006696e-06,
+      "loss": 0.2491,
+      "step": 1574
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.21565376904928e-06,
+      "loss": 0.2396,
+      "step": 1575
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.203557509135774e-06,
+      "loss": 0.2375,
+      "step": 1576
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.191465686459849e-06,
+      "loss": 0.2322,
+      "step": 1577
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1793783202081215e-06,
+      "loss": 0.2452,
+      "step": 1578
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.167295429560145e-06,
+      "loss": 0.408,
+      "step": 1579
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1552170336883735e-06,
+      "loss": 0.1984,
+      "step": 1580
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.143143151758125e-06,
+      "loss": 0.2609,
+      "step": 1581
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.131073802927555e-06,
+      "loss": 0.2316,
+      "step": 1582
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.119009006347625e-06,
+      "loss": 0.332,
+      "step": 1583
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.106948781162079e-06,
+      "loss": 0.2272,
+      "step": 1584
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.094893146507403e-06,
+      "loss": 0.3209,
+      "step": 1585
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0828421215128e-06,
+      "loss": 0.2618,
+      "step": 1586
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.070795725300158e-06,
+      "loss": 0.2091,
+      "step": 1587
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0587539769840185e-06,
+      "loss": 0.2106,
+      "step": 1588
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.04671689567155e-06,
+      "loss": 0.2673,
+      "step": 1589
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.034684500462519e-06,
+      "loss": 0.3133,
+      "step": 1590
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.022656810449253e-06,
+      "loss": 0.2635,
+      "step": 1591
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.01063384471661e-06,
+      "loss": 0.2263,
+      "step": 1592
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.998615622341957e-06,
+      "loss": 0.2539,
+      "step": 1593
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.986602162395133e-06,
+      "loss": 0.2567,
+      "step": 1594
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.974593483938423e-06,
+      "loss": 0.3293,
+      "step": 1595
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.962589606026517e-06,
+      "loss": 0.1816,
+      "step": 1596
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.950590547706497e-06,
+      "loss": 0.2323,
+      "step": 1597
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.938596328017795e-06,
+      "loss": 0.2561,
+      "step": 1598
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.926606965992153e-06,
+      "loss": 0.2479,
+      "step": 1599
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.914622480653629e-06,
+      "loss": 0.2597,
+      "step": 1600
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.902642891018528e-06,
+      "loss": 0.2616,
+      "step": 1601
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.890668216095382e-06,
+      "loss": 0.251,
+      "step": 1602
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.878698474884935e-06,
+      "loss": 0.2604,
+      "step": 1603
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8667336863801e-06,
+      "loss": 0.2292,
+      "step": 1604
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.85477386956593e-06,
+      "loss": 0.28,
+      "step": 1605
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.842819043419585e-06,
+      "loss": 0.2434,
+      "step": 1606
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.830869226910315e-06,
+      "loss": 0.293,
+      "step": 1607
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.818924438999416e-06,
+      "loss": 0.2233,
+      "step": 1608
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.806984698640202e-06,
+      "loss": 0.2923,
+      "step": 1609
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.795050024777984e-06,
+      "loss": 0.2987,
+      "step": 1610
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.78312043635003e-06,
+      "loss": 0.2049,
+      "step": 1611
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.771195952285541e-06,
+      "loss": 0.2257,
+      "step": 1612
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.759276591505614e-06,
+      "loss": 0.3166,
+      "step": 1613
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.747362372923223e-06,
+      "loss": 0.2169,
+      "step": 1614
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.735453315443175e-06,
+      "loss": 0.2357,
+      "step": 1615
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.723549437962098e-06,
+      "loss": 0.2456,
+      "step": 1616
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.7116507593683935e-06,
+      "loss": 0.2465,
+      "step": 1617
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.699757298542214e-06,
+      "loss": 0.2181,
+      "step": 1618
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.687869074355434e-06,
+      "loss": 0.2543,
+      "step": 1619
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.675986105671618e-06,
+      "loss": 0.2247,
+      "step": 1620
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.664108411345993e-06,
+      "loss": 0.351,
+      "step": 1621
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.652236010225417e-06,
+      "loss": 0.218,
+      "step": 1622
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.640368921148345e-06,
+      "loss": 0.196,
+      "step": 1623
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.628507162944808e-06,
+      "loss": 0.346,
+      "step": 1624
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.616650754436376e-06,
+      "loss": 0.2751,
+      "step": 1625
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.604799714436132e-06,
+      "loss": 0.2568,
+      "step": 1626
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.592954061748638e-06,
+      "loss": 0.2863,
+      "step": 1627
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.581113815169908e-06,
+      "loss": 0.324,
+      "step": 1628
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.569278993487382e-06,
+      "loss": 0.2096,
+      "step": 1629
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.5574496154798876e-06,
+      "loss": 0.2833,
+      "step": 1630
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.545625699917609e-06,
+      "loss": 0.266,
+      "step": 1631
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.533807265562085e-06,
+      "loss": 0.258,
+      "step": 1632
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.521994331166131e-06,
+      "loss": 0.246,
+      "step": 1633
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.5101869154738506e-06,
+      "loss": 0.2117,
+      "step": 1634
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.498385037220587e-06,
+      "loss": 0.1647,
+      "step": 1635
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.486588715132893e-06,
+      "loss": 0.2076,
+      "step": 1636
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.474797967928511e-06,
+      "loss": 0.2119,
+      "step": 1637
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.463012814316335e-06,
+      "loss": 0.2409,
+      "step": 1638
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.451233272996386e-06,
+      "loss": 0.3165,
+      "step": 1639
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.439459362659772e-06,
+      "loss": 0.2238,
+      "step": 1640
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.427691101988673e-06,
+      "loss": 0.2738,
+      "step": 1641
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.415928509656306e-06,
+      "loss": 0.2647,
+      "step": 1642
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.404171604326889e-06,
+      "loss": 0.2766,
+      "step": 1643
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.392420404655615e-06,
+      "loss": 0.3304,
+      "step": 1644
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.38067492928863e-06,
+      "loss": 0.1767,
+      "step": 1645
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.3689351968629905e-06,
+      "loss": 0.192,
+      "step": 1646
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.357201226006643e-06,
+      "loss": 0.2006,
+      "step": 1647
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.345473035338397e-06,
+      "loss": 0.2581,
+      "step": 1648
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.3337506434678815e-06,
+      "loss": 0.1932,
+      "step": 1649
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.322034068995528e-06,
+      "loss": 0.2955,
+      "step": 1650
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.31032333051254e-06,
+      "loss": 0.2644,
+      "step": 1651
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.298618446600856e-06,
+      "loss": 0.2564,
+      "step": 1652
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.28691943583313e-06,
+      "loss": 0.2674,
+      "step": 1653
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.275226316772689e-06,
+      "loss": 0.2724,
+      "step": 1654
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.263539107973521e-06,
+      "loss": 0.2479,
+      "step": 1655
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.25185782798023e-06,
+      "loss": 0.2267,
+      "step": 1656
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.240182495328011e-06,
+      "loss": 0.2219,
+      "step": 1657
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.228513128542628e-06,
+      "loss": 0.1774,
+      "step": 1658
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.216849746140375e-06,
+      "loss": 0.3007,
+      "step": 1659
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.20519236662805e-06,
+      "loss": 0.239,
+      "step": 1660
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.1935410085029256e-06,
+      "loss": 0.2273,
+      "step": 1661
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.181895690252724e-06,
+      "loss": 0.238,
+      "step": 1662
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.1702564303555765e-06,
+      "loss": 0.2822,
+      "step": 1663
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.158623247280012e-06,
+      "loss": 0.2911,
+      "step": 1664
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.146996159484907e-06,
+      "loss": 0.2106,
+      "step": 1665
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.135375185419472e-06,
+      "loss": 0.2464,
+      "step": 1666
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.123760343523213e-06,
+      "loss": 0.2714,
+      "step": 1667
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.112151652225906e-06,
+      "loss": 0.2729,
+      "step": 1668
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.1005491299475725e-06,
+      "loss": 0.1999,
+      "step": 1669
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.088952795098442e-06,
+      "loss": 0.2546,
+      "step": 1670
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.077362666078922e-06,
+      "loss": 0.2561,
+      "step": 1671
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.065778761279579e-06,
+      "loss": 0.1796,
+      "step": 1672
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.054201099081103e-06,
+      "loss": 0.1754,
+      "step": 1673
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.042629697854277e-06,
+      "loss": 0.2649,
+      "step": 1674
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.0310645759599455e-06,
+      "loss": 0.2849,
+      "step": 1675
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.019505751748996e-06,
+      "loss": 0.2288,
+      "step": 1676
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.007953243562319e-06,
+      "loss": 0.2702,
+      "step": 1677
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.996407069730786e-06,
+      "loss": 0.212,
+      "step": 1678
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.98486724857521e-06,
+      "loss": 0.2223,
+      "step": 1679
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.973333798406342e-06,
+      "loss": 0.257,
+      "step": 1680
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.961806737524801e-06,
+      "loss": 0.2801,
+      "step": 1681
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.950286084221083e-06,
+      "loss": 0.2065,
+      "step": 1682
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.938771856775511e-06,
+      "loss": 0.2441,
+      "step": 1683
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.927264073458216e-06,
+      "loss": 0.3192,
+      "step": 1684
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.915762752529095e-06,
+      "loss": 0.272,
+      "step": 1685
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.904267912237802e-06,
+      "loss": 0.2568,
+      "step": 1686
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.892779570823702e-06,
+      "loss": 0.246,
+      "step": 1687
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8812977465158465e-06,
+      "loss": 0.2013,
+      "step": 1688
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.869822457532949e-06,
+      "loss": 0.1776,
+      "step": 1689
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.858353722083352e-06,
+      "loss": 0.2053,
+      "step": 1690
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.846891558365002e-06,
+      "loss": 0.2494,
+      "step": 1691
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.835435984565412e-06,
+      "loss": 0.2312,
+      "step": 1692
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8239870188616466e-06,
+      "loss": 0.2219,
+      "step": 1693
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.812544679420273e-06,
+      "loss": 0.2524,
+      "step": 1694
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.801108984397355e-06,
+      "loss": 0.286,
+      "step": 1695
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.789679951938411e-06,
+      "loss": 0.2263,
+      "step": 1696
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.778257600178387e-06,
+      "loss": 0.3123,
+      "step": 1697
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.766841947241628e-06,
+      "loss": 0.2663,
+      "step": 1698
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.755433011241851e-06,
+      "loss": 0.2616,
+      "step": 1699
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.744030810282109e-06,
+      "loss": 0.2841,
+      "step": 1700
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.732635362454775e-06,
+      "loss": 0.2635,
+      "step": 1701
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.721246685841503e-06,
+      "loss": 0.2265,
+      "step": 1702
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.709864798513207e-06,
+      "loss": 0.2636,
+      "step": 1703
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.698489718530022e-06,
+      "loss": 0.23,
+      "step": 1704
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.68712146394129e-06,
+      "loss": 0.2239,
+      "step": 1705
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.675760052785512e-06,
+      "loss": 0.2466,
+      "step": 1706
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.664405503090343e-06,
+      "loss": 0.2209,
+      "step": 1707
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.653057832872535e-06,
+      "loss": 0.2176,
+      "step": 1708
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.641717060137937e-06,
+      "loss": 0.3044,
+      "step": 1709
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.630383202881446e-06,
+      "loss": 0.2378,
+      "step": 1710
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.619056279086988e-06,
+      "loss": 0.3052,
+      "step": 1711
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.607736306727496e-06,
+      "loss": 0.2397,
+      "step": 1712
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.596423303764864e-06,
+      "loss": 0.1892,
+      "step": 1713
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.585117288149922e-06,
+      "loss": 0.3142,
+      "step": 1714
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.573818277822422e-06,
+      "loss": 0.1995,
+      "step": 1715
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.562526290710997e-06,
+      "loss": 0.2641,
+      "step": 1716
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.551241344733138e-06,
+      "loss": 0.268,
+      "step": 1717
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.539963457795163e-06,
+      "loss": 0.2715,
+      "step": 1718
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.528692647792186e-06,
+      "loss": 0.2997,
+      "step": 1719
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.5174289326081e-06,
+      "loss": 0.3045,
+      "step": 1720
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.506172330115527e-06,
+      "loss": 0.1449,
+      "step": 1721
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.49492285817581e-06,
+      "loss": 0.2677,
+      "step": 1722
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.483680534638984e-06,
+      "loss": 0.1747,
+      "step": 1723
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.472445377343731e-06,
+      "loss": 0.2275,
+      "step": 1724
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.4612174041173695e-06,
+      "loss": 0.2397,
+      "step": 1725
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.449996632775813e-06,
+      "loss": 0.2416,
+      "step": 1726
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.438783081123551e-06,
+      "loss": 0.4369,
+      "step": 1727
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.427576766953615e-06,
+      "loss": 0.2598,
+      "step": 1728
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.416377708047555e-06,
+      "loss": 0.2611,
+      "step": 1729
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.4051859221754086e-06,
+      "loss": 0.2405,
+      "step": 1730
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.394001427095671e-06,
+      "loss": 0.2586,
+      "step": 1731
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3828242405552676e-06,
+      "loss": 0.1783,
+      "step": 1732
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.37165438028953e-06,
+      "loss": 0.122,
+      "step": 1733
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.36049186402217e-06,
+      "loss": 0.1377,
+      "step": 1734
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.349336709465232e-06,
+      "loss": 0.0904,
+      "step": 1735
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3381889343190865e-06,
+      "loss": 0.1016,
+      "step": 1736
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.327048556272402e-06,
+      "loss": 0.075,
+      "step": 1737
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.3159155930021e-06,
+      "loss": 0.1013,
+      "step": 1738
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.304790062173342e-06,
+      "loss": 0.1068,
+      "step": 1739
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.293671981439492e-06,
+      "loss": 0.0897,
+      "step": 1740
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.282561368442099e-06,
+      "loss": 0.104,
+      "step": 1741
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.271458240810854e-06,
+      "loss": 0.0683,
+      "step": 1742
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.260362616163573e-06,
+      "loss": 0.0797,
+      "step": 1743
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.249274512106174e-06,
+      "loss": 0.1227,
+      "step": 1744
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.238193946232636e-06,
+      "loss": 0.0865,
+      "step": 1745
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.227120936124976e-06,
+      "loss": 0.1081,
+      "step": 1746
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.216055499353228e-06,
+      "loss": 0.1558,
+      "step": 1747
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.204997653475396e-06,
+      "loss": 0.1199,
+      "step": 1748
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.193947416037453e-06,
+      "loss": 0.0757,
+      "step": 1749
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.182904804573291e-06,
+      "loss": 0.1217,
+      "step": 1750
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.171869836604706e-06,
+      "loss": 0.0674,
+      "step": 1751
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.160842529641366e-06,
+      "loss": 0.0745,
+      "step": 1752
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1498229011807765e-06,
+      "loss": 0.0694,
+      "step": 1753
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.1388109687082676e-06,
+      "loss": 0.076,
+      "step": 1754
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.1278067496969574e-06,
+      "loss": 0.0851,
+      "step": 1755
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.1168102616077145e-06,
+      "loss": 0.0744,
+      "step": 1756
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.105821521889147e-06,
+      "loss": 0.086,
+      "step": 1757
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.094840547977574e-06,
+      "loss": 0.0968,
+      "step": 1758
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.08386735729698e-06,
+      "loss": 0.1049,
+      "step": 1759
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.072901967259014e-06,
+      "loss": 0.0757,
+      "step": 1760
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0619443952629436e-06,
+      "loss": 0.0838,
+      "step": 1761
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.050994658695613e-06,
+      "loss": 0.0878,
+      "step": 1762
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.040052774931459e-06,
+      "loss": 0.0983,
+      "step": 1763
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.029118761332441e-06,
+      "loss": 0.1114,
+      "step": 1764
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.018192635248041e-06,
+      "loss": 0.0804,
+      "step": 1765
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.0072744140152195e-06,
+      "loss": 0.1105,
+      "step": 1766
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.996364114958394e-06,
+      "loss": 0.0982,
+      "step": 1767
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.985461755389418e-06,
+      "loss": 0.0969,
+      "step": 1768
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.974567352607535e-06,
+      "loss": 0.1072,
+      "step": 1769
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9636809238993744e-06,
+      "loss": 0.071,
+      "step": 1770
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.952802486538908e-06,
+      "loss": 0.0671,
+      "step": 1771
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.941932057787428e-06,
+      "loss": 0.0969,
+      "step": 1772
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.931069654893518e-06,
+      "loss": 0.0819,
+      "step": 1773
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.920215295093028e-06,
+      "loss": 0.1002,
+      "step": 1774
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.9093689956090474e-06,
+      "loss": 0.0784,
+      "step": 1775
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.898530773651872e-06,
+      "loss": 0.0852,
+      "step": 1776
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.887700646418982e-06,
+      "loss": 0.0705,
+      "step": 1777
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.876878631095017e-06,
+      "loss": 0.0847,
+      "step": 1778
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.866064744851737e-06,
+      "loss": 0.0793,
+      "step": 1779
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.855259004848013e-06,
+      "loss": 0.1197,
+      "step": 1780
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.844461428229782e-06,
+      "loss": 0.0633,
+      "step": 1781
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.833672032130036e-06,
+      "loss": 0.0887,
+      "step": 1782
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.822890833668773e-06,
+      "loss": 0.0816,
+      "step": 1783
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.812117849952997e-06,
+      "loss": 0.0838,
+      "step": 1784
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.801353098076671e-06,
+      "loss": 0.0936,
+      "step": 1785
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.790596595120699e-06,
+      "loss": 0.1116,
+      "step": 1786
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7798483581528935e-06,
+      "loss": 0.0634,
+      "step": 1787
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.769108404227952e-06,
+      "loss": 0.0706,
+      "step": 1788
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.75837675038743e-06,
+      "loss": 0.1048,
+      "step": 1789
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.747653413659716e-06,
+      "loss": 0.1169,
+      "step": 1790
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.736938411059983e-06,
+      "loss": 0.0845,
+      "step": 1791
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.726231759590209e-06,
+      "loss": 0.0745,
+      "step": 1792
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.7155334762391e-06,
+      "loss": 0.0986,
+      "step": 1793
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.704843577982091e-06,
+      "loss": 0.0685,
+      "step": 1794
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.694162081781311e-06,
+      "loss": 0.0737,
+      "step": 1795
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6834890045855594e-06,
+      "loss": 0.1033,
+      "step": 1796
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.672824363330269e-06,
+      "loss": 0.0765,
+      "step": 1797
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.662168174937496e-06,
+      "loss": 0.0951,
+      "step": 1798
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.65152045631588e-06,
+      "loss": 0.0807,
+      "step": 1799
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6408812243606204e-06,
+      "loss": 0.0757,
+      "step": 1800
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.630250495953453e-06,
+      "loss": 0.0858,
+      "step": 1801
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.619628287962619e-06,
+      "loss": 0.0992,
+      "step": 1802
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6090146172428454e-06,
+      "loss": 0.0803,
+      "step": 1803
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.598409500635297e-06,
+      "loss": 0.0989,
+      "step": 1804
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.587812954967583e-06,
+      "loss": 0.0758,
+      "step": 1805
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.577224997053704e-06,
+      "loss": 0.0974,
+      "step": 1806
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.5666456436940335e-06,
+      "loss": 0.094,
+      "step": 1807
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.5560749116753e-06,
+      "loss": 0.0948,
+      "step": 1808
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.545512817770547e-06,
+      "loss": 0.1048,
+      "step": 1809
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.534959378739104e-06,
+      "loss": 0.0677,
+      "step": 1810
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.524414611326575e-06,
+      "loss": 0.1111,
+      "step": 1811
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.513878532264806e-06,
+      "loss": 0.0885,
+      "step": 1812
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.503351158271854e-06,
+      "loss": 0.0849,
+      "step": 1813
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.492832506051964e-06,
+      "loss": 0.088,
+      "step": 1814
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.482322592295541e-06,
+      "loss": 0.1077,
+      "step": 1815
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.471821433679124e-06,
+      "loss": 0.1262,
+      "step": 1816
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.461329046865365e-06,
+      "loss": 0.093,
+      "step": 1817
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.4508454485029854e-06,
+      "loss": 0.1002,
+      "step": 1818
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.440370655226773e-06,
+      "loss": 0.0865,
+      "step": 1819
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.42990468365754e-06,
+      "loss": 0.111,
+      "step": 1820
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.4194475504021e-06,
+      "loss": 0.0893,
+      "step": 1821
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.408999272053247e-06,
+      "loss": 0.0872,
+      "step": 1822
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.398559865189718e-06,
+      "loss": 0.0998,
+      "step": 1823
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.388129346376177e-06,
+      "loss": 0.0898,
+      "step": 1824
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.377707732163184e-06,
+      "loss": 0.0792,
+      "step": 1825
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.3672950390871715e-06,
+      "loss": 0.0852,
+      "step": 1826
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.356891283670414e-06,
+      "loss": 0.07,
+      "step": 1827
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.346496482421005e-06,
+      "loss": 0.0986,
+      "step": 1828
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.33611065183283e-06,
+      "loss": 0.0948,
+      "step": 1829
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.325733808385544e-06,
+      "loss": 0.0818,
+      "step": 1830
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.315365968544532e-06,
+      "loss": 0.0823,
+      "step": 1831
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.305007148760901e-06,
+      "loss": 0.0869,
+      "step": 1832
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.294657365471445e-06,
+      "loss": 0.0783,
+      "step": 1833
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.284316635098617e-06,
+      "loss": 0.1094,
+      "step": 1834
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.273984974050504e-06,
+      "loss": 0.0962,
+      "step": 1835
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.263662398720807e-06,
+      "loss": 0.0753,
+      "step": 1836
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.253348925488807e-06,
+      "loss": 0.0793,
+      "step": 1837
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2430445707193465e-06,
+      "loss": 0.0923,
+      "step": 1838
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.232749350762785e-06,
+      "loss": 0.1148,
+      "step": 1839
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.222463281955008e-06,
+      "loss": 0.1,
+      "step": 1840
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.212186380617369e-06,
+      "loss": 0.0527,
+      "step": 1841
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.201918663056674e-06,
+      "loss": 0.1078,
+      "step": 1842
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.191660145565162e-06,
+      "loss": 0.0629,
+      "step": 1843
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.181410844420473e-06,
+      "loss": 0.0949,
+      "step": 1844
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.171170775885616e-06,
+      "loss": 0.0711,
+      "step": 1845
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.160939956208958e-06,
+      "loss": 0.0899,
+      "step": 1846
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.150718401624186e-06,
+      "loss": 0.0893,
+      "step": 1847
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.140506128350292e-06,
+      "loss": 0.0819,
+      "step": 1848
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.130303152591533e-06,
+      "loss": 0.0992,
+      "step": 1849
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.120109490537417e-06,
+      "loss": 0.075,
+      "step": 1850
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.1099251583626786e-06,
+      "loss": 0.0808,
+      "step": 1851
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.099750172227235e-06,
+      "loss": 0.0832,
+      "step": 1852
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.089584548276185e-06,
+      "loss": 0.0995,
+      "step": 1853
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.079428302639769e-06,
+      "loss": 0.0971,
+      "step": 1854
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.069281451433342e-06,
+      "loss": 0.0768,
+      "step": 1855
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0591440107573635e-06,
+      "loss": 0.0952,
+      "step": 1856
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.049015996697355e-06,
+      "loss": 0.0749,
+      "step": 1857
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.038897425323871e-06,
+      "loss": 0.0939,
+      "step": 1858
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.0287883126924946e-06,
+      "loss": 0.0609,
+      "step": 1859
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.0186886748437985e-06,
+      "loss": 0.1191,
+      "step": 1860
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.008598527803317e-06,
+      "loss": 0.0643,
+      "step": 1861
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.998517887581529e-06,
+      "loss": 0.0693,
+      "step": 1862
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.988446770173828e-06,
+      "loss": 0.0627,
+      "step": 1863
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.978385191560495e-06,
+      "loss": 0.0773,
+      "step": 1864
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9683331677066825e-06,
+      "loss": 0.0663,
+      "step": 1865
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.958290714562366e-06,
+      "loss": 0.0868,
+      "step": 1866
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.948257848062351e-06,
+      "loss": 0.0711,
+      "step": 1867
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.938234584126224e-06,
+      "loss": 0.0911,
+      "step": 1868
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.928220938658337e-06,
+      "loss": 0.0719,
+      "step": 1869
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.918216927547779e-06,
+      "loss": 0.0811,
+      "step": 1870
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9082225666683505e-06,
+      "loss": 0.1102,
+      "step": 1871
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.898237871878543e-06,
+      "loss": 0.0679,
+      "step": 1872
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.888262859021508e-06,
+      "loss": 0.0825,
+      "step": 1873
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8782975439250335e-06,
+      "loss": 0.0778,
+      "step": 1874
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.868341942401521e-06,
+      "loss": 0.0742,
+      "step": 1875
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.858396070247959e-06,
+      "loss": 0.0796,
+      "step": 1876
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.848459943245898e-06,
+      "loss": 0.0967,
+      "step": 1877
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.838533577161428e-06,
+      "loss": 0.0824,
+      "step": 1878
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.828616987745142e-06,
+      "loss": 0.0919,
+      "step": 1879
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.818710190732128e-06,
+      "loss": 0.0908,
+      "step": 1880
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.8088132018419353e-06,
+      "loss": 0.0787,
+      "step": 1881
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7989260367785497e-06,
+      "loss": 0.0773,
+      "step": 1882
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7890487112303665e-06,
+      "loss": 0.0937,
+      "step": 1883
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7791812408701712e-06,
+      "loss": 0.0945,
+      "step": 1884
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7693236413551094e-06,
+      "loss": 0.091,
+      "step": 1885
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.75947592832667e-06,
+      "loss": 0.0993,
+      "step": 1886
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7496381174106388e-06,
+      "loss": 0.0779,
+      "step": 1887
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7398102242171106e-06,
+      "loss": 0.0797,
+      "step": 1888
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7299922643404308e-06,
+      "loss": 0.0769,
+      "step": 1889
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7201842533591847e-06,
+      "loss": 0.0885,
+      "step": 1890
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.710386206836173e-06,
+      "loss": 0.112,
+      "step": 1891
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7005981403183868e-06,
+      "loss": 0.0957,
+      "step": 1892
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.690820069336973e-06,
+      "loss": 0.0861,
+      "step": 1893
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.681052009407229e-06,
+      "loss": 0.0935,
+      "step": 1894
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6712939760285603e-06,
+      "loss": 0.0689,
+      "step": 1895
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.661545984684467e-06,
+      "loss": 0.0771,
+      "step": 1896
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6518080508425124e-06,
+      "loss": 0.0859,
+      "step": 1897
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6420801899543023e-06,
+      "loss": 0.0628,
+      "step": 1898
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.632362417455458e-06,
+      "loss": 0.133,
+      "step": 1899
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6226547487655994e-06,
+      "loss": 0.0767,
+      "step": 1900
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6129571992882993e-06,
+      "loss": 0.1125,
+      "step": 1901
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6032697844110896e-06,
+      "loss": 0.097,
+      "step": 1902
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.59359251950541e-06,
+      "loss": 0.0558,
+      "step": 1903
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.583925419926607e-06,
+      "loss": 0.0806,
+      "step": 1904
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5742685010138877e-06,
+      "loss": 0.0953,
+      "step": 1905
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5646217780903104e-06,
+      "loss": 0.0846,
+      "step": 1906
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.554985266462744e-06,
+      "loss": 0.0826,
+      "step": 1907
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.545358981421868e-06,
+      "loss": 0.0838,
+      "step": 1908
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5357429382421294e-06,
+      "loss": 0.0719,
+      "step": 1909
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5261371521817247e-06,
+      "loss": 0.0907,
+      "step": 1910
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5165416384825735e-06,
+      "loss": 0.0663,
+      "step": 1911
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.506956412370297e-06,
+      "loss": 0.0802,
+      "step": 1912
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.497381489054198e-06,
+      "loss": 0.0715,
+      "step": 1913
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4878168837272174e-06,
+      "loss": 0.0925,
+      "step": 1914
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4782626115659355e-06,
+      "loss": 0.1111,
+      "step": 1915
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4687186877305367e-06,
+      "loss": 0.0751,
+      "step": 1916
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.459185127364779e-06,
+      "loss": 0.0877,
+      "step": 1917
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4496619455959814e-06,
+      "loss": 0.0915,
+      "step": 1918
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.440149157534992e-06,
+      "loss": 0.0983,
+      "step": 1919
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4306467782761678e-06,
+      "loss": 0.1079,
+      "step": 1920
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.421154822897349e-06,
+      "loss": 0.0843,
+      "step": 1921
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4116733064598375e-06,
+      "loss": 0.0826,
+      "step": 1922
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.402202244008369e-06,
+      "loss": 0.0952,
+      "step": 1923
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.392741650571092e-06,
+      "loss": 0.069,
+      "step": 1924
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3832915411595446e-06,
+      "loss": 0.0783,
+      "step": 1925
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.373851930768629e-06,
+      "loss": 0.0765,
+      "step": 1926
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3644228343765904e-06,
+      "loss": 0.1017,
+      "step": 1927
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.355004266944981e-06,
+      "loss": 0.0747,
+      "step": 1928
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3455962434186594e-06,
+      "loss": 0.0895,
+      "step": 1929
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3361987787257445e-06,
+      "loss": 0.1248,
+      "step": 1930
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.326811887777607e-06,
+      "loss": 0.0877,
+      "step": 1931
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3174355854688366e-06,
+      "loss": 0.0878,
+      "step": 1932
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.308069886677221e-06,
+      "loss": 0.0923,
+      "step": 1933
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2987148062637288e-06,
+      "loss": 0.0781,
+      "step": 1934
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2893703590724645e-06,
+      "loss": 0.0806,
+      "step": 1935
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2800365599306818e-06,
+      "loss": 0.0859,
+      "step": 1936
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2707134236487226e-06,
+      "loss": 0.0853,
+      "step": 1937
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2614009650200173e-06,
+      "loss": 0.097,
+      "step": 1938
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2520991988210494e-06,
+      "loss": 0.0732,
+      "step": 1939
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2428081398113433e-06,
+      "loss": 0.0618,
+      "step": 1940
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2335278027334195e-06,
+      "loss": 0.0931,
+      "step": 1941
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2242582023127987e-06,
+      "loss": 0.0733,
+      "step": 1942
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2149993532579637e-06,
+      "loss": 0.0758,
+      "step": 1943
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.205751270260332e-06,
+      "loss": 0.0755,
+      "step": 1944
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1965139679942424e-06,
+      "loss": 0.1063,
+      "step": 1945
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.187287461116926e-06,
+      "loss": 0.0915,
+      "step": 1946
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1780717642684855e-06,
+      "loss": 0.0774,
+      "step": 1947
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1688668920718733e-06,
+      "loss": 0.0787,
+      "step": 1948
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1596728591328553e-06,
+      "loss": 0.1041,
+      "step": 1949
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1504896800400087e-06,
+      "loss": 0.0583,
+      "step": 1950
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1413173693646825e-06,
+      "loss": 0.097,
+      "step": 1951
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.13215594166099e-06,
+      "loss": 0.0882,
+      "step": 1952
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.123005411465766e-06,
+      "loss": 0.0887,
+      "step": 1953
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.113865793298558e-06,
+      "loss": 0.0775,
+      "step": 1954
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1047371016615945e-06,
+      "loss": 0.0744,
+      "step": 1955
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0956193510397712e-06,
+      "loss": 0.0661,
+      "step": 1956
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0865125559006217e-06,
+      "loss": 0.0808,
+      "step": 1957
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.077416730694296e-06,
+      "loss": 0.1126,
+      "step": 1958
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0683318898535376e-06,
+      "loss": 0.0784,
+      "step": 1959
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0592580477936606e-06,
+      "loss": 0.0806,
+      "step": 1960
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0501952189125317e-06,
+      "loss": 0.0768,
+      "step": 1961
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0411434175905296e-06,
+      "loss": 0.0828,
+      "step": 1962
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0321026581905467e-06,
+      "loss": 0.0883,
+      "step": 1963
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.023072955057952e-06,
+      "loss": 0.0635,
+      "step": 1964
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0140543225205697e-06,
+      "loss": 0.0893,
+      "step": 1965
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.005046774888659e-06,
+      "loss": 0.0801,
+      "step": 1966
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.996050326454889e-06,
+      "loss": 0.0834,
+      "step": 1967
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9870649914943173e-06,
+      "loss": 0.0827,
+      "step": 1968
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.978090784264369e-06,
+      "loss": 0.0945,
+      "step": 1969
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9691277190048107e-06,
+      "loss": 0.1024,
+      "step": 1970
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9601758099377297e-06,
+      "loss": 0.0795,
+      "step": 1971
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.951235071267511e-06,
+      "loss": 0.0625,
+      "step": 1972
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9423055171808167e-06,
+      "loss": 0.0815,
+      "step": 1973
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9333871618465604e-06,
+      "loss": 0.0879,
+      "step": 1974
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9244800194158895e-06,
+      "loss": 0.0917,
+      "step": 1975
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.91558410402215e-06,
+      "loss": 0.089,
+      "step": 1976
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9066994297808826e-06,
+      "loss": 0.0854,
+      "step": 1977
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8978260107897893e-06,
+      "loss": 0.0942,
+      "step": 1978
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.888963861128712e-06,
+      "loss": 0.0825,
+      "step": 1979
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8801129948596098e-06,
+      "loss": 0.0736,
+      "step": 1980
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.871273426026541e-06,
+      "loss": 0.0751,
+      "step": 1981
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.862445168655633e-06,
+      "loss": 0.0747,
+      "step": 1982
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8536282367550715e-06,
+      "loss": 0.1061,
+      "step": 1983
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.844822644315066e-06,
+      "loss": 0.0747,
+      "step": 1984
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8360284053078347e-06,
+      "loss": 0.0768,
+      "step": 1985
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8272455336875825e-06,
+      "loss": 0.0741,
+      "step": 1986
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8184740433904743e-06,
+      "loss": 0.08,
+      "step": 1987
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.809713948334618e-06,
+      "loss": 0.0672,
+      "step": 1988
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.8009652624200436e-06,
+      "loss": 0.0937,
+      "step": 1989
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7922279995286648e-06,
+      "loss": 0.0852,
+      "step": 1990
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7835021735242853e-06,
+      "loss": 0.0988,
+      "step": 1991
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.774787798252552e-06,
+      "loss": 0.0743,
+      "step": 1992
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7660848875409453e-06,
+      "loss": 0.1002,
+      "step": 1993
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7573934551987547e-06,
+      "loss": 0.0835,
+      "step": 1994
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7487135150170563e-06,
+      "loss": 0.0943,
+      "step": 1995
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.740045080768694e-06,
+      "loss": 0.0837,
+      "step": 1996
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7313881662082455e-06,
+      "loss": 0.1106,
+      "step": 1997
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.722742785072018e-06,
+      "loss": 0.0839,
+      "step": 1998
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7141089510780117e-06,
+      "loss": 0.1201,
+      "step": 1999
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.705486677925917e-06,
+      "loss": 0.0854,
+      "step": 2000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6968759792970665e-06,
+      "loss": 0.0916,
+      "step": 2001
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.688276868854437e-06,
+      "loss": 0.0651,
+      "step": 2002
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6796893602426054e-06,
+      "loss": 0.0968,
+      "step": 2003
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6711134670877513e-06,
+      "loss": 0.0681,
+      "step": 2004
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6625492029976184e-06,
+      "loss": 0.0772,
+      "step": 2005
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.653996581561501e-06,
+      "loss": 0.0846,
+      "step": 2006
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6454556163502154e-06,
+      "loss": 0.0829,
+      "step": 2007
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6369263209160846e-06,
+      "loss": 0.0761,
+      "step": 2008
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6284087087929155e-06,
+      "loss": 0.1186,
+      "step": 2009
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.619902793495979e-06,
+      "loss": 0.1366,
+      "step": 2010
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6114085885219763e-06,
+      "loss": 0.0783,
+      "step": 2011
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6029261073490365e-06,
+      "loss": 0.07,
+      "step": 2012
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.594455363436683e-06,
+      "loss": 0.0713,
+      "step": 2013
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5859963702258174e-06,
+      "loss": 0.0665,
+      "step": 2014
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.57754914113869e-06,
+      "loss": 0.0897,
+      "step": 2015
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.569113689578897e-06,
+      "loss": 0.1087,
+      "step": 2016
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.56069002893133e-06,
+      "loss": 0.089,
+      "step": 2017
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5522781725621814e-06,
+      "loss": 0.103,
+      "step": 2018
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5438781338189145e-06,
+      "loss": 0.1148,
+      "step": 2019
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5354899260302346e-06,
+      "loss": 0.0835,
+      "step": 2020
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5271135625060795e-06,
+      "loss": 0.082,
+      "step": 2021
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5187490565375917e-06,
+      "loss": 0.0857,
+      "step": 2022
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.510396421397101e-06,
+      "loss": 0.0602,
+      "step": 2023
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.502055670338093e-06,
+      "loss": 0.0966,
+      "step": 2024
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4937268165952044e-06,
+      "loss": 0.0866,
+      "step": 2025
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4854098733841935e-06,
+      "loss": 0.0867,
+      "step": 2026
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.477104853901915e-06,
+      "loss": 0.0839,
+      "step": 2027
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4688117713263093e-06,
+      "loss": 0.0991,
+      "step": 2028
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4605306388163707e-06,
+      "loss": 0.0846,
+      "step": 2029
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.452261469512136e-06,
+      "loss": 0.1031,
+      "step": 2030
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.444004276534657e-06,
+      "loss": 0.0793,
+      "step": 2031
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.435759072985984e-06,
+      "loss": 0.0775,
+      "step": 2032
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.427525871949139e-06,
+      "loss": 0.071,
+      "step": 2033
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.419304686488104e-06,
+      "loss": 0.0942,
+      "step": 2034
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4110955296477954e-06,
+      "loss": 0.0701,
+      "step": 2035
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.402898414454038e-06,
+      "loss": 0.0737,
+      "step": 2036
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.394713353913557e-06,
+      "loss": 0.082,
+      "step": 2037
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.38654036101394e-06,
+      "loss": 0.0618,
+      "step": 2038
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3783794487236367e-06,
+      "loss": 0.0829,
+      "step": 2039
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3702306299919207e-06,
+      "loss": 0.1294,
+      "step": 2040
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.362093917748881e-06,
+      "loss": 0.0955,
+      "step": 2041
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3539693249053943e-06,
+      "loss": 0.0956,
+      "step": 2042
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3458568643531078e-06,
+      "loss": 0.1048,
+      "step": 2043
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3377565489644184e-06,
+      "loss": 0.0725,
+      "step": 2044
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3296683915924455e-06,
+      "loss": 0.1126,
+      "step": 2045
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.321592405071027e-06,
+      "loss": 0.0645,
+      "step": 2046
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3135286022146785e-06,
+      "loss": 0.0707,
+      "step": 2047
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3054769958185962e-06,
+      "loss": 0.0724,
+      "step": 2048
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2974375986586118e-06,
+      "loss": 0.0921,
+      "step": 2049
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2894104234911917e-06,
+      "loss": 0.0769,
+      "step": 2050
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2813954830534e-06,
+      "loss": 0.0686,
+      "step": 2051
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.273392790062896e-06,
+      "loss": 0.104,
+      "step": 2052
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.265402357217903e-06,
+      "loss": 0.0725,
+      "step": 2053
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.257424197197189e-06,
+      "loss": 0.0748,
+      "step": 2054
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.249458322660051e-06,
+      "loss": 0.0833,
+      "step": 2055
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.241504746246289e-06,
+      "loss": 0.0994,
+      "step": 2056
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.233563480576191e-06,
+      "loss": 0.0789,
+      "step": 2057
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.225634538250514e-06,
+      "loss": 0.0873,
+      "step": 2058
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2177179318504506e-06,
+      "loss": 0.0897,
+      "step": 2059
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.20981367393763e-06,
+      "loss": 0.1085,
+      "step": 2060
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2019217770540824e-06,
+      "loss": 0.1055,
+      "step": 2061
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1940422537222285e-06,
+      "loss": 0.0919,
+      "step": 2062
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1861751164448476e-06,
+      "loss": 0.0954,
+      "step": 2063
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1783203777050788e-06,
+      "loss": 0.0945,
+      "step": 2064
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1704780499663715e-06,
+      "loss": 0.0697,
+      "step": 2065
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1626481456724937e-06,
+      "loss": 0.0631,
+      "step": 2066
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1548306772474957e-06,
+      "loss": 0.0737,
+      "step": 2067
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.147025657095697e-06,
+      "loss": 0.0694,
+      "step": 2068
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.139233097601664e-06,
+      "loss": 0.0739,
+      "step": 2069
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1314530111301925e-06,
+      "loss": 0.0808,
+      "step": 2070
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1236854100262838e-06,
+      "loss": 0.0861,
+      "step": 2071
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.115930306615136e-06,
+      "loss": 0.0726,
+      "step": 2072
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1081877132021023e-06,
+      "loss": 0.1087,
+      "step": 2073
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1004576420726995e-06,
+      "loss": 0.081,
+      "step": 2074
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.092740105492568e-06,
+      "loss": 0.0907,
+      "step": 2075
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.08503511570746e-06,
+      "loss": 0.0878,
+      "step": 2076
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.077342684943221e-06,
+      "loss": 0.1041,
+      "step": 2077
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0696628254057673e-06,
+      "loss": 0.1034,
+      "step": 2078
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0619955492810673e-06,
+      "loss": 0.1056,
+      "step": 2079
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.054340868735123e-06,
+      "loss": 0.0616,
+      "step": 2080
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0466987959139505e-06,
+      "loss": 0.0942,
+      "step": 2081
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0390693429435626e-06,
+      "loss": 0.085,
+      "step": 2082
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0314525219299443e-06,
+      "loss": 0.0971,
+      "step": 2083
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0238483449590387e-06,
+      "loss": 0.0849,
+      "step": 2084
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0162568240967284e-06,
+      "loss": 0.0553,
+      "step": 2085
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.008677971388808e-06,
+      "loss": 0.0692,
+      "step": 2086
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.001111798860975e-06,
+      "loss": 0.0836,
+      "step": 2087
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.993558318518807e-06,
+      "loss": 0.0751,
+      "step": 2088
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9860175423477414e-06,
+      "loss": 0.0933,
+      "step": 2089
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9784894823130587e-06,
+      "loss": 0.1044,
+      "step": 2090
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.970974150359859e-06,
+      "loss": 0.1012,
+      "step": 2091
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9634715584130483e-06,
+      "loss": 0.0991,
+      "step": 2092
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.955981718377321e-06,
+      "loss": 0.1142,
+      "step": 2093
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.948504642137129e-06,
+      "loss": 0.0763,
+      "step": 2094
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9410403415566737e-06,
+      "loss": 0.0778,
+      "step": 2095
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9335888284798955e-06,
+      "loss": 0.0764,
+      "step": 2096
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9261501147304308e-06,
+      "loss": 0.0687,
+      "step": 2097
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.918724212111613e-06,
+      "loss": 0.0869,
+      "step": 2098
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9113111324064504e-06,
+      "loss": 0.0637,
+      "step": 2099
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9039108873775924e-06,
+      "loss": 0.0701,
+      "step": 2100
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8965234887673367e-06,
+      "loss": 0.0753,
+      "step": 2101
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8891489482975912e-06,
+      "loss": 0.1308,
+      "step": 2102
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.881787277669861e-06,
+      "loss": 0.0701,
+      "step": 2103
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.874438488565231e-06,
+      "loss": 0.0789,
+      "step": 2104
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8671025926443464e-06,
+      "loss": 0.0841,
+      "step": 2105
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8597796015473967e-06,
+      "loss": 0.0828,
+      "step": 2106
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8524695268940863e-06,
+      "loss": 0.0956,
+      "step": 2107
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.845172380283633e-06,
+      "loss": 0.0895,
+      "step": 2108
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8378881732947395e-06,
+      "loss": 0.0886,
+      "step": 2109
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8306169174855747e-06,
+      "loss": 0.0978,
+      "step": 2110
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8233586243937552e-06,
+      "loss": 0.075,
+      "step": 2111
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.81611330553634e-06,
+      "loss": 0.0811,
+      "step": 2112
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8088809724097856e-06,
+      "loss": 0.0917,
+      "step": 2113
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8016616364899541e-06,
+      "loss": 0.0703,
+      "step": 2114
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7944553092320816e-06,
+      "loss": 0.0509,
+      "step": 2115
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.787262002070762e-06,
+      "loss": 0.0743,
+      "step": 2116
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.780081726419931e-06,
+      "loss": 0.0931,
+      "step": 2117
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7729144936728459e-06,
+      "loss": 0.0954,
+      "step": 2118
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7657603152020676e-06,
+      "loss": 0.0761,
+      "step": 2119
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7586192023594462e-06,
+      "loss": 0.0907,
+      "step": 2120
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7514911664760937e-06,
+      "loss": 0.0856,
+      "step": 2121
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7443762188623792e-06,
+      "loss": 0.0769,
+      "step": 2122
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7372743708078999e-06,
+      "loss": 0.074,
+      "step": 2123
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7301856335814683e-06,
+      "loss": 0.075,
+      "step": 2124
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7231100184310955e-06,
+      "loss": 0.0805,
+      "step": 2125
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.716047536583968e-06,
+      "loss": 0.1007,
+      "step": 2126
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7089981992464355e-06,
+      "loss": 0.0775,
+      "step": 2127
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7019620176039886e-06,
+      "loss": 0.0732,
+      "step": 2128
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6949390028212453e-06,
+      "loss": 0.0671,
+      "step": 2129
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6879291660419306e-06,
+      "loss": 0.08,
+      "step": 2130
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.680932518388858e-06,
+      "loss": 0.074,
+      "step": 2131
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.673949070963915e-06,
+      "loss": 0.0889,
+      "step": 2132
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6669788348480453e-06,
+      "loss": 0.1066,
+      "step": 2133
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.660021821101222e-06,
+      "loss": 0.0621,
+      "step": 2134
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6530780407624458e-06,
+      "loss": 0.0704,
+      "step": 2135
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6461475048497167e-06,
+      "loss": 0.086,
+      "step": 2136
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.639230224360019e-06,
+      "loss": 0.0705,
+      "step": 2137
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6323262102693038e-06,
+      "loss": 0.0854,
+      "step": 2138
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6254354735324728e-06,
+      "loss": 0.0799,
+      "step": 2139
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6185580250833577e-06,
+      "loss": 0.0967,
+      "step": 2140
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6116938758347112e-06,
+      "loss": 0.0553,
+      "step": 2141
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.604843036678173e-06,
+      "loss": 0.0818,
+      "step": 2142
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5980055184842702e-06,
+      "loss": 0.0802,
+      "step": 2143
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.591181332102395e-06,
+      "loss": 0.0881,
+      "step": 2144
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5843704883607813e-06,
+      "loss": 0.0913,
+      "step": 2145
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5775729980664921e-06,
+      "loss": 0.0728,
+      "step": 2146
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5707888720054042e-06,
+      "loss": 0.0876,
+      "step": 2147
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5640181209421845e-06,
+      "loss": 0.0874,
+      "step": 2148
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.55726075562028e-06,
+      "loss": 0.0828,
+      "step": 2149
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5505167867618986e-06,
+      "loss": 0.087,
+      "step": 2150
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5437862250679903e-06,
+      "loss": 0.059,
+      "step": 2151
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5370690812182309e-06,
+      "loss": 0.0565,
+      "step": 2152
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5303653658710082e-06,
+      "loss": 0.0947,
+      "step": 2153
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5236750896634011e-06,
+      "loss": 0.0932,
+      "step": 2154
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5169982632111602e-06,
+      "loss": 0.0952,
+      "step": 2155
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5103348971086985e-06,
+      "loss": 0.0845,
+      "step": 2156
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5036850019290728e-06,
+      "loss": 0.0946,
+      "step": 2157
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4970485882239594e-06,
+      "loss": 0.105,
+      "step": 2158
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.490425666523646e-06,
+      "loss": 0.09,
+      "step": 2159
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4838162473370188e-06,
+      "loss": 0.1167,
+      "step": 2160
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.477220341151524e-06,
+      "loss": 0.0977,
+      "step": 2161
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4706379584331788e-06,
+      "loss": 0.0587,
+      "step": 2162
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4640691096265358e-06,
+      "loss": 0.1039,
+      "step": 2163
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4575138051546756e-06,
+      "loss": 0.0996,
+      "step": 2164
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4509720554191864e-06,
+      "loss": 0.0752,
+      "step": 2165
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4444438708001473e-06,
+      "loss": 0.08,
+      "step": 2166
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4379292616561157e-06,
+      "loss": 0.0693,
+      "step": 2167
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4314282383241097e-06,
+      "loss": 0.068,
+      "step": 2168
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4249408111195805e-06,
+      "loss": 0.1003,
+      "step": 2169
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4184669903364167e-06,
+      "loss": 0.077,
+      "step": 2170
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4120067862469112e-06,
+      "loss": 0.0599,
+      "step": 2171
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.405560209101754e-06,
+      "loss": 0.0951,
+      "step": 2172
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3991272691300096e-06,
+      "loss": 0.0834,
+      "step": 2173
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3927079765391049e-06,
+      "loss": 0.0705,
+      "step": 2174
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3863023415148146e-06,
+      "loss": 0.0767,
+      "step": 2175
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3799103742212372e-06,
+      "loss": 0.0742,
+      "step": 2176
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.373532084800787e-06,
+      "loss": 0.0986,
+      "step": 2177
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3671674833741765e-06,
+      "loss": 0.0659,
+      "step": 2178
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.360816580040396e-06,
+      "loss": 0.079,
+      "step": 2179
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3544793848767012e-06,
+      "loss": 0.0761,
+      "step": 2180
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3481559079385965e-06,
+      "loss": 0.1083,
+      "step": 2181
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3418461592598232e-06,
+      "loss": 0.0846,
+      "step": 2182
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3355501488523292e-06,
+      "loss": 0.0697,
+      "step": 2183
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3292678867062714e-06,
+      "loss": 0.0715,
+      "step": 2184
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3229993827899922e-06,
+      "loss": 0.1046,
+      "step": 2185
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.316744647049999e-06,
+      "loss": 0.0943,
+      "step": 2186
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3105036894109547e-06,
+      "loss": 0.0799,
+      "step": 2187
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3042765197756603e-06,
+      "loss": 0.0964,
+      "step": 2188
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2980631480250394e-06,
+      "loss": 0.0834,
+      "step": 2189
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.291863584018117e-06,
+      "loss": 0.081,
+      "step": 2190
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.285677837592012e-06,
+      "loss": 0.0638,
+      "step": 2191
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.279505918561923e-06,
+      "loss": 0.0814,
+      "step": 2192
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2733478367211027e-06,
+      "loss": 0.0958,
+      "step": 2193
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2672036018408485e-06,
+      "loss": 0.1015,
+      "step": 2194
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.261073223670487e-06,
+      "loss": 0.0715,
+      "step": 2195
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.254956711937355e-06,
+      "loss": 0.0779,
+      "step": 2196
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2488540763467904e-06,
+      "loss": 0.0748,
+      "step": 2197
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2427653265821115e-06,
+      "loss": 0.1118,
+      "step": 2198
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2366904723046048e-06,
+      "loss": 0.0672,
+      "step": 2199
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2306295231535058e-06,
+      "loss": 0.0934,
+      "step": 2200
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.22458248874599e-06,
+      "loss": 0.0769,
+      "step": 2201
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2185493786771485e-06,
+      "loss": 0.0951,
+      "step": 2202
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2125302025199847e-06,
+      "loss": 0.0873,
+      "step": 2203
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2065249698253845e-06,
+      "loss": 0.065,
+      "step": 2204
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2005336901221131e-06,
+      "loss": 0.0866,
+      "step": 2205
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1945563729167998e-06,
+      "loss": 0.1031,
+      "step": 2206
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1885930276939095e-06,
+      "loss": 0.0703,
+      "step": 2207
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1826436639157479e-06,
+      "loss": 0.0652,
+      "step": 2208
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1767082910224316e-06,
+      "loss": 0.1009,
+      "step": 2209
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.170786918431872e-06,
+      "loss": 0.0929,
+      "step": 2210
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1648795555397719e-06,
+      "loss": 0.0914,
+      "step": 2211
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1589862117196005e-06,
+      "loss": 0.0833,
+      "step": 2212
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1531068963225866e-06,
+      "loss": 0.0978,
+      "step": 2213
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1472416186776969e-06,
+      "loss": 0.0922,
+      "step": 2214
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1413903880916221e-06,
+      "loss": 0.0865,
+      "step": 2215
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1355532138487691e-06,
+      "loss": 0.061,
+      "step": 2216
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.129730105211232e-06,
+      "loss": 0.0766,
+      "step": 2217
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1239210714187954e-06,
+      "loss": 0.0954,
+      "step": 2218
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1181261216889073e-06,
+      "loss": 0.0688,
+      "step": 2219
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1123452652166667e-06,
+      "loss": 0.0638,
+      "step": 2220
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1065785111748117e-06,
+      "loss": 0.0792,
+      "step": 2221
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.1008258687137042e-06,
+      "loss": 0.1047,
+      "step": 2222
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.095087346961311e-06,
+      "loss": 0.0801,
+      "step": 2223
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0893629550231988e-06,
+      "loss": 0.0875,
+      "step": 2224
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0836527019825071e-06,
+      "loss": 0.0916,
+      "step": 2225
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0779565968999473e-06,
+      "loss": 0.0787,
+      "step": 2226
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.072274648813777e-06,
+      "loss": 0.0877,
+      "step": 2227
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0666068667397912e-06,
+      "loss": 0.0858,
+      "step": 2228
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.060953259671308e-06,
+      "loss": 0.1057,
+      "step": 2229
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0553138365791538e-06,
+      "loss": 0.0915,
+      "step": 2230
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0496886064116462e-06,
+      "loss": 0.0806,
+      "step": 2231
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0440775780945844e-06,
+      "loss": 0.0997,
+      "step": 2232
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.038480760531233e-06,
+      "loss": 0.082,
+      "step": 2233
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.032898162602306e-06,
+      "loss": 0.0918,
+      "step": 2234
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0273297931659565e-06,
+      "loss": 0.0805,
+      "step": 2235
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.02177566105776e-06,
+      "loss": 0.0781,
+      "step": 2236
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0162357750907015e-06,
+      "loss": 0.0777,
+      "step": 2237
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.010710144055158e-06,
+      "loss": 0.0908,
+      "step": 2238
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.0051987767188886e-06,
+      "loss": 0.0943,
+      "step": 2239
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.997016818270256e-07,
+      "loss": 0.075,
+      "step": 2240
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.942188681020482e-07,
+      "loss": 0.0824,
+      "step": 2241
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.88750344243775e-07,
+      "loss": 0.0704,
+      "step": 2242
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.832961189293555e-07,
+      "loss": 0.1072,
+      "step": 2243
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.77856200813243e-07,
+      "loss": 0.0694,
+      "step": 2244
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.724305985271952e-07,
+      "loss": 0.0655,
+      "step": 2245
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.670193206802526e-07,
+      "loss": 0.0843,
+      "step": 2246
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.616223758587262e-07,
+      "loss": 0.0707,
+      "step": 2247
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.56239772626185e-07,
+      "loss": 0.0859,
+      "step": 2248
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.508715195234397e-07,
+      "loss": 0.0829,
+      "step": 2249
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.455176250685338e-07,
+      "loss": 0.0684,
+      "step": 2250
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.401780977567265e-07,
+      "loss": 0.0806,
+      "step": 2251
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.348529460604761e-07,
+      "loss": 0.0596,
+      "step": 2252
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.295421784294345e-07,
+      "loss": 0.0814,
+      "step": 2253
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.242458032904311e-07,
+      "loss": 0.1008,
+      "step": 2254
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.189638290474512e-07,
+      "loss": 0.0785,
+      "step": 2255
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.136962640816405e-07,
+      "loss": 0.0699,
+      "step": 2256
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.084431167512731e-07,
+      "loss": 0.0548,
+      "step": 2257
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.032043953917436e-07,
+      "loss": 0.0667,
+      "step": 2258
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.979801083155637e-07,
+      "loss": 0.0921,
+      "step": 2259
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.92770263812337e-07,
+      "loss": 0.1041,
+      "step": 2260
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.875748701487519e-07,
+      "loss": 0.0643,
+      "step": 2261
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.823939355685662e-07,
+      "loss": 0.0891,
+      "step": 2262
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.772274682925963e-07,
+      "loss": 0.1007,
+      "step": 2263
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.720754765187011e-07,
+      "loss": 0.0762,
+      "step": 2264
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.669379684217738e-07,
+      "loss": 0.055,
+      "step": 2265
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.618149521537178e-07,
+      "loss": 0.1062,
+      "step": 2266
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.567064358434507e-07,
+      "loss": 0.0985,
+      "step": 2267
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.516124275968773e-07,
+      "loss": 0.0743,
+      "step": 2268
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.465329354968844e-07,
+      "loss": 0.0989,
+      "step": 2269
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.414679676033233e-07,
+      "loss": 0.0654,
+      "step": 2270
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.36417531953001e-07,
+      "loss": 0.0817,
+      "step": 2271
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.313816365596627e-07,
+      "loss": 0.0969,
+      "step": 2272
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.263602894139855e-07,
+      "loss": 0.0888,
+      "step": 2273
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.213534984835603e-07,
+      "loss": 0.0724,
+      "step": 2274
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.163612717128811e-07,
+      "loss": 0.0814,
+      "step": 2275
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.113836170233324e-07,
+      "loss": 0.0912,
+      "step": 2276
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.064205423131766e-07,
+      "loss": 0.0654,
+      "step": 2277
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.014720554575417e-07,
+      "loss": 0.0887,
+      "step": 2278
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.965381643084069e-07,
+      "loss": 0.0781,
+      "step": 2279
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.916188766945931e-07,
+      "loss": 0.0793,
+      "step": 2280
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.867142004217499e-07,
+      "loss": 0.0808,
+      "step": 2281
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.818241432723417e-07,
+      "loss": 0.0949,
+      "step": 2282
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.769487130056352e-07,
+      "loss": 0.055,
+      "step": 2283
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.7208791735769e-07,
+      "loss": 0.0885,
+      "step": 2284
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.67241764041341e-07,
+      "loss": 0.0712,
+      "step": 2285
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.624102607461958e-07,
+      "loss": 0.0919,
+      "step": 2286
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.575934151386055e-07,
+      "loss": 0.0874,
+      "step": 2287
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.527912348616751e-07,
+      "loss": 0.0658,
+      "step": 2288
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.480037275352314e-07,
+      "loss": 0.1009,
+      "step": 2289
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.432309007558236e-07,
+      "loss": 0.073,
+      "step": 2290
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.384727620967025e-07,
+      "loss": 0.0806,
+      "step": 2291
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.337293191078177e-07,
+      "loss": 0.0812,
+      "step": 2292
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.290005793157939e-07,
+      "loss": 0.0765,
+      "step": 2293
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.242865502239304e-07,
+      "loss": 0.0716,
+      "step": 2294
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.195872393121849e-07,
+      "loss": 0.0771,
+      "step": 2295
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.149026540371573e-07,
+      "loss": 0.0907,
+      "step": 2296
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.102328018320859e-07,
+      "loss": 0.1079,
+      "step": 2297
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.055776901068278e-07,
+      "loss": 0.0973,
+      "step": 2298
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.009373262478558e-07,
+      "loss": 0.0897,
+      "step": 2299
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.963117176182332e-07,
+      "loss": 0.1123,
+      "step": 2300
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.917008715576168e-07,
+      "loss": 0.0762,
+      "step": 2301
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.871047953822396e-07,
+      "loss": 0.0524,
+      "step": 2302
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.825234963848937e-07,
+      "loss": 0.0851,
+      "step": 2303
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.779569818349296e-07,
+      "loss": 0.0744,
+      "step": 2304
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.73405258978237e-07,
+      "loss": 0.0729,
+      "step": 2305
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.688683350372282e-07,
+      "loss": 0.1158,
+      "step": 2306
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.643462172108417e-07,
+      "loss": 0.087,
+      "step": 2307
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.598389126745209e-07,
+      "loss": 0.0721,
+      "step": 2308
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.553464285802003e-07,
+      "loss": 0.0624,
+      "step": 2309
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.508687720563023e-07,
+      "loss": 0.0752,
+      "step": 2310
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.464059502077203e-07,
+      "loss": 0.101,
+      "step": 2311
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.419579701158085e-07,
+      "loss": 0.0812,
+      "step": 2312
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.375248388383726e-07,
+      "loss": 0.0901,
+      "step": 2313
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.331065634096545e-07,
+      "loss": 0.0669,
+      "step": 2314
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.28703150840324e-07,
+      "loss": 0.0811,
+      "step": 2315
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.243146081174711e-07,
+      "loss": 0.0774,
+      "step": 2316
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.199409422045854e-07,
+      "loss": 0.0994,
+      "step": 2317
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.155821600415568e-07,
+      "loss": 0.0672,
+      "step": 2318
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.112382685446538e-07,
+      "loss": 0.0871,
+      "step": 2319
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.069092746065197e-07,
+      "loss": 0.0632,
+      "step": 2320
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.025951850961576e-07,
+      "loss": 0.0702,
+      "step": 2321
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.982960068589227e-07,
+      "loss": 0.0738,
+      "step": 2322
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.940117467165096e-07,
+      "loss": 0.0909,
+      "step": 2323
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.897424114669392e-07,
+      "loss": 0.074,
+      "step": 2324
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.854880078845549e-07,
+      "loss": 0.0711,
+      "step": 2325
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.812485427200043e-07,
+      "loss": 0.0797,
+      "step": 2326
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.770240227002311e-07,
+      "loss": 0.0754,
+      "step": 2327
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.728144545284653e-07,
+      "loss": 0.087,
+      "step": 2328
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.686198448842151e-07,
+      "loss": 0.0928,
+      "step": 2329
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.644402004232496e-07,
+      "loss": 0.0793,
+      "step": 2330
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.602755277775929e-07,
+      "loss": 0.0748,
+      "step": 2331
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.561258335555142e-07,
+      "loss": 0.0948,
+      "step": 2332
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.519911243415143e-07,
+      "loss": 0.0656,
+      "step": 2333
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.478714066963198e-07,
+      "loss": 0.1026,
+      "step": 2334
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.437666871568603e-07,
+      "loss": 0.0774,
+      "step": 2335
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.396769722362805e-07,
+      "loss": 0.0863,
+      "step": 2336
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.35602268423906e-07,
+      "loss": 0.0687,
+      "step": 2337
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.315425821852494e-07,
+      "loss": 0.0878,
+      "step": 2338
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.274979199619912e-07,
+      "loss": 0.0903,
+      "step": 2339
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.234682881719766e-07,
+      "loss": 0.0662,
+      "step": 2340
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.194536932091943e-07,
+      "loss": 0.0643,
+      "step": 2341
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.154541414437797e-07,
+      "loss": 0.0802,
+      "step": 2342
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.114696392219953e-07,
+      "loss": 0.0906,
+      "step": 2343
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.075001928662271e-07,
+      "loss": 0.0774,
+      "step": 2344
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.03545808674969e-07,
+      "loss": 0.0783,
+      "step": 2345
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.996064929228128e-07,
+      "loss": 0.0907,
+      "step": 2346
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.956822518604476e-07,
+      "loss": 0.08,
+      "step": 2347
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.917730917146335e-07,
+      "loss": 0.0762,
+      "step": 2348
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.878790186882076e-07,
+      "loss": 0.1126,
+      "step": 2349
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.840000389600663e-07,
+      "loss": 0.0958,
+      "step": 2350
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.801361586851539e-07,
+      "loss": 0.0695,
+      "step": 2351
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.762873839944626e-07,
+      "loss": 0.0713,
+      "step": 2352
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.7245372099500973e-07,
+      "loss": 0.0921,
+      "step": 2353
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6863517576983485e-07,
+      "loss": 0.0747,
+      "step": 2354
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6483175437799123e-07,
+      "loss": 0.0782,
+      "step": 2355
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.610434628545335e-07,
+      "loss": 0.0807,
+      "step": 2356
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5727030721051005e-07,
+      "loss": 0.0658,
+      "step": 2357
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.535122934329528e-07,
+      "loss": 0.0907,
+      "step": 2358
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.497694274848641e-07,
+      "loss": 0.0805,
+      "step": 2359
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.460417153052155e-07,
+      "loss": 0.0721,
+      "step": 2360
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.423291628089299e-07,
+      "loss": 0.0816,
+      "step": 2361
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3863177588687636e-07,
+      "loss": 0.0733,
+      "step": 2362
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3494956040586066e-07,
+      "loss": 0.0856,
+      "step": 2363
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3128252220861476e-07,
+      "loss": 0.0956,
+      "step": 2364
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2763066711378867e-07,
+      "loss": 0.057,
+      "step": 2365
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2399400091594154e-07,
+      "loss": 0.0844,
+      "step": 2366
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2037252938552875e-07,
+      "loss": 0.0868,
+      "step": 2367
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1676625826889916e-07,
+      "loss": 0.0975,
+      "step": 2368
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.131751932882799e-07,
+      "loss": 0.0566,
+      "step": 2369
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0959934014177193e-07,
+      "loss": 0.1062,
+      "step": 2370
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.060387045033365e-07,
+      "loss": 0.0697,
+      "step": 2371
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0249329202278977e-07,
+      "loss": 0.0681,
+      "step": 2372
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.98963108325795e-07,
+      "loss": 0.064,
+      "step": 2373
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.954481590138459e-07,
+      "loss": 0.091,
+      "step": 2374
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9194844966427e-07,
+      "loss": 0.094,
+      "step": 2375
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.884639858302053e-07,
+      "loss": 0.0673,
+      "step": 2376
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.849947730406045e-07,
+      "loss": 0.0767,
+      "step": 2377
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.815408168002188e-07,
+      "loss": 0.0962,
+      "step": 2378
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.781021225895931e-07,
+      "loss": 0.1073,
+      "step": 2379
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7467869586505166e-07,
+      "loss": 0.0851,
+      "step": 2380
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.712705420586971e-07,
+      "loss": 0.084,
+      "step": 2381
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6787766657839675e-07,
+      "loss": 0.0742,
+      "step": 2382
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.645000748077709e-07,
+      "loss": 0.0696,
+      "step": 2383
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6113777210619483e-07,
+      "loss": 0.0956,
+      "step": 2384
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.577907638087819e-07,
+      "loss": 0.1015,
+      "step": 2385
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.5445905522637493e-07,
+      "loss": 0.0932,
+      "step": 2386
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.5114265164554185e-07,
+      "loss": 0.0703,
+      "step": 2387
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.478415583285677e-07,
+      "loss": 0.0587,
+      "step": 2388
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4455578051343696e-07,
+      "loss": 0.0773,
+      "step": 2389
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.412853234138402e-07,
+      "loss": 0.0814,
+      "step": 2390
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3803019221915424e-07,
+      "loss": 0.0726,
+      "step": 2391
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3479039209443733e-07,
+      "loss": 0.0621,
+      "step": 2392
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.315659281804229e-07,
+      "loss": 0.0884,
+      "step": 2393
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2835680559350716e-07,
+      "loss": 0.0761,
+      "step": 2394
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.2516302942574794e-07,
+      "loss": 0.0833,
+      "step": 2395
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.2198460474484827e-07,
+      "loss": 0.0789,
+      "step": 2396
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.188215365941527e-07,
+      "loss": 0.0795,
+      "step": 2397
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1567382999263986e-07,
+      "loss": 0.0858,
+      "step": 2398
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1254148993491327e-07,
+      "loss": 0.0699,
+      "step": 2399
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0942452139119504e-07,
+      "loss": 0.0794,
+      "step": 2400
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0632292930731556e-07,
+      "loss": 0.085,
+      "step": 2401
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0323671860470695e-07,
+      "loss": 0.0728,
+      "step": 2402
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.001658941803931e-07,
+      "loss": 0.0809,
+      "step": 2403
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.971104609069864e-07,
+      "loss": 0.0742,
+      "step": 2404
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9407042363267526e-07,
+      "loss": 0.0859,
+      "step": 2405
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9104578718122113e-07,
+      "loss": 0.0574,
+      "step": 2406
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8803655635194603e-07,
+      "loss": 0.0781,
+      "step": 2407
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.850427359197283e-07,
+      "loss": 0.0788,
+      "step": 2408
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8206433063499573e-07,
+      "loss": 0.0697,
+      "step": 2409
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7910134522371126e-07,
+      "loss": 0.1003,
+      "step": 2410
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7615378438737406e-07,
+      "loss": 0.0774,
+      "step": 2411
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.7322165280300737e-07,
+      "loss": 0.0794,
+      "step": 2412
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.7030495512315516e-07,
+      "loss": 0.0825,
+      "step": 2413
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.674036959758664e-07,
+      "loss": 0.0681,
+      "step": 2414
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6451787996469744e-07,
+      "loss": 0.0629,
+      "step": 2415
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.616475116686978e-07,
+      "loss": 0.1229,
+      "step": 2416
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5879259564240646e-07,
+      "loss": 0.0743,
+      "step": 2417
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5595313641584406e-07,
+      "loss": 0.1226,
+      "step": 2418
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.531291384945056e-07,
+      "loss": 0.0862,
+      "step": 2419
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5032060635935e-07,
+      "loss": 0.0694,
+      "step": 2420
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4752754446679793e-07,
+      "loss": 0.0766,
+      "step": 2421
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4474995724872553e-07,
+      "loss": 0.0933,
+      "step": 2422
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4198784911245055e-07,
+      "loss": 0.0665,
+      "step": 2423
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.392412244407294e-07,
+      "loss": 0.0795,
+      "step": 2424
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3651008759175253e-07,
+      "loss": 0.0687,
+      "step": 2425
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3379444289913344e-07,
+      "loss": 0.0684,
+      "step": 2426
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3109429467190637e-07,
+      "loss": 0.0876,
+      "step": 2427
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2840964719451186e-07,
+      "loss": 0.0574,
+      "step": 2428
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.257405047268002e-07,
+      "loss": 0.1214,
+      "step": 2429
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2308687150401465e-07,
+      "loss": 0.0738,
+      "step": 2430
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2044875173679146e-07,
+      "loss": 0.0839,
+      "step": 2431
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1782614961115001e-07,
+      "loss": 0.0839,
+      "step": 2432
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1521906928848923e-07,
+      "loss": 0.1015,
+      "step": 2433
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1262751490557677e-07,
+      "loss": 0.0698,
+      "step": 2434
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1005149057454434e-07,
+      "loss": 0.0811,
+      "step": 2435
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0749100038288561e-07,
+      "loss": 0.0888,
+      "step": 2436
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0494604839343847e-07,
+      "loss": 0.0767,
+      "step": 2437
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0241663864439044e-07,
+      "loss": 0.0909,
+      "step": 2438
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.999027751492677e-07,
+      "loss": 0.0483,
+      "step": 2439
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.974044618969262e-07,
+      "loss": 0.0781,
+      "step": 2440
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9492170285155042e-07,
+      "loss": 0.0788,
+      "step": 2441
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9245450195264025e-07,
+      "loss": 0.0707,
+      "step": 2442
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9000286311501415e-07,
+      "loss": 0.0935,
+      "step": 2443
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8756679022879364e-07,
+      "loss": 0.0772,
+      "step": 2444
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8514628715940008e-07,
+      "loss": 0.1084,
+      "step": 2445
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8274135774755452e-07,
+      "loss": 0.0767,
+      "step": 2446
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8035200580926003e-07,
+      "loss": 0.0933,
+      "step": 2447
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7797823513580946e-07,
+      "loss": 0.0829,
+      "step": 2448
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7562004949376767e-07,
+      "loss": 0.0981,
+      "step": 2449
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.732774526249703e-07,
+      "loss": 0.0894,
+      "step": 2450
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7095044824651741e-07,
+      "loss": 0.0914,
+      "step": 2451
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.686390400507698e-07,
+      "loss": 0.0941,
+      "step": 2452
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6634323170533928e-07,
+      "loss": 0.0943,
+      "step": 2453
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6406302685308518e-07,
+      "loss": 0.0894,
+      "step": 2454
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6179842911210997e-07,
+      "loss": 0.0793,
+      "step": 2455
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5954944207574707e-07,
+      "loss": 0.0784,
+      "step": 2456
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5731606931256526e-07,
+      "loss": 0.0722,
+      "step": 2457
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5509831436635425e-07,
+      "loss": 0.071,
+      "step": 2458
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5289618075612023e-07,
+      "loss": 0.101,
+      "step": 2459
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.507096719760881e-07,
+      "loss": 0.1025,
+      "step": 2460
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4853879149568483e-07,
+      "loss": 0.095,
+      "step": 2461
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.463835427595428e-07,
+      "loss": 0.0859,
+      "step": 2462
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4424392918748864e-07,
+      "loss": 0.1011,
+      "step": 2463
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4211995417454106e-07,
+      "loss": 0.0792,
+      "step": 2464
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4001162109090416e-07,
+      "loss": 0.0755,
+      "step": 2465
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3791893328196083e-07,
+      "loss": 0.0733,
+      "step": 2466
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3584189406827265e-07,
+      "loss": 0.0755,
+      "step": 2467
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3378050674556775e-07,
+      "loss": 0.1114,
+      "step": 2468
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.317347745847386e-07,
+      "loss": 0.0588,
+      "step": 2469
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.297047008318386e-07,
+      "loss": 0.0884,
+      "step": 2470
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2769028870807443e-07,
+      "loss": 0.103,
+      "step": 2471
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2569154140980255e-07,
+      "loss": 0.0906,
+      "step": 2472
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2370846210852162e-07,
+      "loss": 0.0709,
+      "step": 2473
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.217410539508701e-07,
+      "loss": 0.1149,
+      "step": 2474
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1978932005862088e-07,
+      "loss": 0.1024,
+      "step": 2475
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1785326352867665e-07,
+      "loss": 0.0941,
+      "step": 2476
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1593288743306231e-07,
+      "loss": 0.0887,
+      "step": 2477
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1402819481892147e-07,
+      "loss": 0.0883,
+      "step": 2478
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1213918870851437e-07,
+      "loss": 0.1014,
+      "step": 2479
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1026587209920892e-07,
+      "loss": 0.0712,
+      "step": 2480
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0840824796347848e-07,
+      "loss": 0.0694,
+      "step": 2481
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0656631924889749e-07,
+      "loss": 0.085,
+      "step": 2482
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.047400888781347e-07,
+      "loss": 0.0817,
+      "step": 2483
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.029295597489488e-07,
+      "loss": 0.0734,
+      "step": 2484
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0113473473418512e-07,
+      "loss": 0.0718,
+      "step": 2485
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.935561668177218e-08,
+      "loss": 0.0714,
+      "step": 2486
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.759220841471406e-08,
+      "loss": 0.0656,
+      "step": 2487
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.584451273108809e-08,
+      "loss": 0.0754,
+      "step": 2488
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.411253240403817e-08,
+      "loss": 0.0762,
+      "step": 2489
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.239627018177378e-08,
+      "loss": 0.0663,
+      "step": 2490
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.069572878756427e-08,
+      "loss": 0.091,
+      "step": 2491
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.901091091973235e-08,
+      "loss": 0.0798,
+      "step": 2492
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.734181925165175e-08,
+      "loss": 0.0578,
+      "step": 2493
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.568845643174284e-08,
+      "loss": 0.0611,
+      "step": 2494
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.405082508346929e-08,
+      "loss": 0.0741,
+      "step": 2495
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.242892780533362e-08,
+      "loss": 0.0777,
+      "step": 2496
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.082276717086945e-08,
+      "loss": 0.0729,
+      "step": 2497
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.923234572864369e-08,
+      "loss": 0.0883,
+      "step": 2498
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.765766600224544e-08,
+      "loss": 0.0888,
+      "step": 2499
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.609873049028826e-08,
+      "loss": 0.0827,
+      "step": 2500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.455554166640566e-08,
+      "loss": 0.0615,
+      "step": 2501
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.302810197924004e-08,
+      "loss": 0.0534,
+      "step": 2502
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.151641385244823e-08,
+      "loss": 0.0743,
+      "step": 2503
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.00204796846915e-08,
+      "loss": 0.0851,
+      "step": 2504
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.854030184963334e-08,
+      "loss": 0.0854,
+      "step": 2505
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.707588269593834e-08,
+      "loss": 0.0909,
+      "step": 2506
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.562722454726333e-08,
+      "loss": 0.0864,
+      "step": 2507
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.419432970225736e-08,
+      "loss": 0.0735,
+      "step": 2508
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.277720043455726e-08,
+      "loss": 0.1145,
+      "step": 2509
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.137583899278543e-08,
+      "loss": 0.0671,
+      "step": 2510
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.999024760054095e-08,
+      "loss": 0.1009,
+      "step": 2511
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.862042845640403e-08,
+      "loss": 0.1022,
+      "step": 2512
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.7266383733926015e-08,
+      "loss": 0.0815,
+      "step": 2513
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.592811558162936e-08,
+      "loss": 0.0569,
+      "step": 2514
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.4605626123002134e-08,
+      "loss": 0.0509,
+      "step": 2515
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.329891745649574e-08,
+      "loss": 0.0968,
+      "step": 2516
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.200799165552273e-08,
+      "loss": 0.101,
+      "step": 2517
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.073285076845236e-08,
+      "loss": 0.0761,
+      "step": 2518
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.947349681860503e-08,
+      "loss": 0.0972,
+      "step": 2519
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.822993180425228e-08,
+      "loss": 0.1039,
+      "step": 2520
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.7002157698614603e-08,
+      "loss": 0.07,
+      "step": 2521
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.579017644985362e-08,
+      "loss": 0.0894,
+      "step": 2522
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.459398998107434e-08,
+      "loss": 0.0916,
+      "step": 2523
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.3413600190316265e-08,
+      "loss": 0.0688,
+      "step": 2524
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.2249008950557836e-08,
+      "loss": 0.0747,
+      "step": 2525
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.1100218109704214e-08,
+      "loss": 0.0819,
+      "step": 2526
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9967229490595064e-08,
+      "loss": 0.0633,
+      "step": 2527
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8850044890990094e-08,
+      "loss": 0.0936,
+      "step": 2528
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7748666083576856e-08,
+      "loss": 0.0629,
+      "step": 2529
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.666309481595964e-08,
+      "loss": 0.081,
+      "step": 2530
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5593332810663907e-08,
+      "loss": 0.0924,
+      "step": 2531
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.453938176512628e-08,
+      "loss": 0.0784,
+      "step": 2532
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.350124335169791e-08,
+      "loss": 0.0685,
+      "step": 2533
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.2478919217636685e-08,
+      "loss": 0.0757,
+      "step": 2534
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.147241098511167e-08,
+      "loss": 0.0646,
+      "step": 2535
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0481720251190894e-08,
+      "loss": 0.0836,
+      "step": 2536
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9506848587848024e-08,
+      "loss": 0.086,
+      "step": 2537
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.854779754195569e-08,
+      "loss": 0.0747,
+      "step": 2538
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.760456863527883e-08,
+      "loss": 0.0938,
+      "step": 2539
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.667716336448356e-08,
+      "loss": 0.0673,
+      "step": 2540
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5765583201121657e-08,
+      "loss": 0.0643,
+      "step": 2541
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4869829591639417e-08,
+      "loss": 0.0764,
+      "step": 2542
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3989903957366556e-08,
+      "loss": 0.0997,
+      "step": 2543
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3125807694519553e-08,
+      "loss": 0.0836,
+      "step": 2544
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.227754217419942e-08,
+      "loss": 0.0688,
+      "step": 2545
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1445108742383925e-08,
+      "loss": 0.0671,
+      "step": 2546
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.062850871993427e-08,
+      "loss": 0.072,
+      "step": 2547
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9827743402582866e-08,
+      "loss": 0.0877,
+      "step": 2548
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9042814060939995e-08,
+      "loss": 0.1017,
+      "step": 2549
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8273721940488265e-08,
+      "loss": 0.0899,
+      "step": 2550
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.752046826157816e-08,
+      "loss": 0.0817,
+      "step": 2551
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6783054219431382e-08,
+      "loss": 0.0975,
+      "step": 2552
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.606148098413418e-08,
+      "loss": 0.1003,
+      "step": 2553
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5355749700639576e-08,
+      "loss": 0.119,
+      "step": 2554
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4665861488761813e-08,
+      "loss": 0.0861,
+      "step": 2555
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3991817443175238e-08,
+      "loss": 0.0824,
+      "step": 2556
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3333618633417645e-08,
+      "loss": 0.1,
+      "step": 2557
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2691266103879162e-08,
+      "loss": 0.0809,
+      "step": 2558
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2064760873810033e-08,
+      "loss": 0.0965,
+      "step": 2559
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1454103937313943e-08,
+      "loss": 0.0761,
+      "step": 2560
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0859296263345809e-08,
+      "loss": 0.0914,
+      "step": 2561
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0280338795713996e-08,
+      "loss": 0.0825,
+      "step": 2562
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.717232453074765e-09,
+      "loss": 0.1129,
+      "step": 2563
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.169978128934498e-09,
+      "loss": 0.0751,
+      "step": 2564
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.63857669164414e-09,
+      "loss": 0.0803,
+      "step": 2565
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.123028984403647e-09,
+      "loss": 0.0756,
+      "step": 2566
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.623335825253097e-09,
+      "loss": 0.0626,
+      "step": 2567
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.139498007078249e-09,
+      "loss": 0.0725,
+      "step": 2568
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.671516297606095e-09,
+      "loss": 0.0688,
+      "step": 2569
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.219391439403754e-09,
+      "loss": 0.0817,
+      "step": 2570
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.7831241498762516e-09,
+      "loss": 0.1034,
+      "step": 2571
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.362715121269846e-09,
+      "loss": 0.0764,
+      "step": 2572
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.958165020664263e-09,
+      "loss": 0.0938,
+      "step": 2573
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.569474489978243e-09,
+      "loss": 0.0582,
+      "step": 2574
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.19664414596177e-09,
+      "loss": 0.0784,
+      "step": 2575
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.8396745802038446e-09,
+      "loss": 0.0732,
+      "step": 2576
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.4985663591224906e-09,
+      "loss": 0.0964,
+      "step": 2577
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.173320023968085e-09,
+      "loss": 0.0839,
+      "step": 2578
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.863936090825581e-09,
+      "loss": 0.0912,
+      "step": 2579
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5704150506045135e-09,
+      "loss": 0.0726,
+      "step": 2580
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.292757369051213e-09,
+      "loss": 0.0649,
+      "step": 2581
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0309634867354823e-09,
+      "loss": 0.0629,
+      "step": 2582
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7850338190561479e-09,
+      "loss": 0.0761,
+      "step": 2583
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.55496875624217e-09,
+      "loss": 0.0755,
+      "step": 2584
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.340768663348202e-09,
+      "loss": 0.0744,
+      "step": 2585
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1424338802534796e-09,
+      "loss": 0.0585,
+      "step": 2586
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.599647216640417e-10,
+      "loss": 0.0689,
+      "step": 2587
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.933614771138409e-10,
+      "loss": 0.072,
+      "step": 2588
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.426244109580815e-10,
+      "loss": 0.0969,
+      "step": 2589
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.077537623776607e-10,
+      "loss": 0.0675,
+      "step": 2590
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.8874974537916887e-10,
+      "loss": 0.0858,
+      "step": 2591
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.85612548789338e-10,
+      "loss": 0.0741,
+      "step": 2592
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9834233626281342e-10,
+      "loss": 0.0857,
+      "step": 2593
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.2693924627327214e-10,
+      "loss": 0.0815,
+      "step": 2594
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.140339212008407e-11,
+      "loss": 0.0949,
+      "step": 2595
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.1734861923871097e-11,
+      "loss": 0.0795,
+      "step": 2596
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.933718627617381e-12,
+      "loss": 0.0897,
+      "step": 2597
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.0482,
+      "step": 2598
+    },
+    {
+      "epoch": 3.0,
+      "step": 2598,
+      "total_flos": 2.0249694493646782e+19,
+      "train_loss": 0.30513740509455944,
+      "train_runtime": 39405.9312,
+      "train_samples_per_second": 3.162,
+      "train_steps_per_second": 0.066
+    }
+  ],
+  "max_steps": 2598,
+  "num_train_epochs": 3,
+  "total_flos": 2.0249694493646782e+19,
+  "trial_name": null,
+  "trial_params": null
+}